×
Apple beta season is here

Apple beta season is here

Hi, friends! Welcome to Installer No. 91, your guide to the best and Verge-iest stuff in the world. (If you’re new here, welcome, how is summer more than a month over already, and also you can read all the old editions at the Installer homepage.)

This week, I’m pondering the leaked Pixel 10 lineup, marveling at BYD’s leaping Yangwang U9 supercar, reading about why everyone’s reading Reddit, already looking forward to Ted Lasso’s fourth season, rewatching MKBHD’s videos about the Escobar folding phone, melting with joy at the new Pokémon Concierge season 2 trailer, dreaming of buying the Lego Game Boy but not actually doing so because my toddler would destroy it, and listening to Lucius’ recently released self-titled album.

I also have for you some new betas from Apple, a retro-styled PC, some thoughts about cases, and how to play Dance Dance Revolution at home today.

(As always, the best part of Installer is your ideas and tips. What do you want to know more about? What awesome tricks do you know that everyone else should? What app should everyone be using? Tell me everything: installer@theverge.com. And if you know someone else who might enjoy Installer, forward it to them and tell them to subscribe here.)

  • Apple’s public betas: Public betas for iOS 26, iPadOS 26, macOS Tahoe 26, watchOS 26, and tvOS 26 are available now, giving you a chance to see Apple’s Liquid Glass design language in action before the updates are released widely this fall. I’ve been running the developer betas, and after some necessary tweaks from Apple, I’m warming up to Liquid Glass more than I expected.
  • Maingear’s Retro95 PC: This limited release puts a retro shell over modern PC components. Anyone have $1,599 (or more) they can lend me?
  • AppleCare One: This new $19.99-per-month subscription from Apple lets you cover up to three devices at once with the same benefits you’d get with individual AppleCare Plus subscriptions. You can add more devices for $5.99 per month each. I don’t usually pay for warranty coverage, but AppleCare One could be a good deal for a lot of people.
  • The Fantastic Four: First Steps: It’s Marvel’s turn to release a superhero movie with some pretty high stakes for the company that makes it — and like DC and Superman, Marvel appears to have mostly pulled it off. As a huge Michael Giacchino fan, about half the reason I want to see this movie is for the score.
  • Amazon’s Kindle Colorsoft (16GB): This new version of Amazon’s color Kindle comes with half the storage and is missing a few perks from the Kindle Colorsoft Signature Edition, but at $249.99, it’s $30 cheaper than last year’s model. I’m going to hold on to my 10th-generation Kindle Paperwhite as long as I can, but whenever it’s time to upgrade that, I’m sorely tempted by a color ereader.
  • Roomba Max 705 Combo: This new combo robot vacuum from iRobot has a “first-of-its-kind roller mop cover that keeps carpets dry.” If that cover actually works, it seems like this could be a great dual-purpose robovacuum.
  • Wheel World: It’s a cycling-themed open-world adventure game from Annapurna Interactive and the developers of Nidhogg. That sounds like a winning combo to me, and it might just be the feel-good game of the summer.
  • Acapulco season 4: I watched the first two seasons of this Apple TV Plus series set at a resort in Mexico and loved them — the show is funny and heartwarming. Like Ted Lasso, it can be a little saccharine, but I think we should all watch some lighter TV every once in a while. Might be time for me to catch up.

Thanks to everyone who replied to my quandary about whether or not I should keep using a phone case. There were a bunch of great perspectives and feelings, so I’m sharing a few of them here in the newsletter — a lot more people go caseless than I expected!

As for my own decision? Well, I haven’t reached one yet, and honestly, I may never pick one lane. Typically, my iPhone 16 Pro lives in this slim Torras case, but as I write this, it’s sitting on top of the case, not in it. Will I put the case on before I walk out the door? We’ll see!

“I, too, felt the urge to go caseless. I’m pretty careful with my electronics, so I still didn’t love the idea of setting my phone down on tables and things alike. I got the magnetic back from Nomad, and it has felt like the perfect meet-in-the-middle accessory for me! Phone feels almost caseless, and I still get the back protection.” – Omesh

“I am totally the same with cases. I know I’ll drop it once a year and totally ruin the thing and I hate that. What I do is take the case off when I’m just chilling at my desk or couch or chair. That way I get the joy of the smooth phone, but when I am in danger of dropping it on concrete or getting out of my truck, I have the protection it needs.” – Travis

“I ran my first gen Pixel Fold without a case (while living in a beachy resort town) for two years! If the Fold can survive it, so can a 16 Pro. The Fold is still alive and well as a secondary device/backup phone. Of course, it’s definitely not pristine, so if you don’t want your phone to bear its scars, you should probably keep a case on it.” – John

“I have the Peak Design GNAR case. Saw some complaints online about the protection being too minimal for the front of the phone, but it’s been solid for me. Does well being outdoors (I run outside all the time). I’m rocking no screen protector.” – airwr3ck6669

“I never used cases until I got my current phone (Pixel 6A) because the Razer Kishi controller I used no longer fit snuggly around it, so I found a case that allowed it to fit again. I would say the case would eventually come off, but I only realized as I was reading this article that even though I no longer use that controller, I hadn’t thought to take the case off.” – xPutNameHerex

Today, I’m featuring Keith Broni, the editor-in-chief of Emojipedia. I thought of reaching out to him for this section after World Emoji Day on July 17th (the date shown on the calendar emoji), and he graciously replied. He initially got involved with the site in 2018 and took on increasing responsibilities “over the years” before taking the editor-in-chief job in 2022.

In addition to everything about his homescreen, I wanted to know what he’s looking forward to in the world of emoji. “I’m excited to see what the next unexpected emoji repurposing meme is going to be, especially as Gen Alpha becomes more and more online,” he tells me, like how Millennials repurposed the Eggplant and Peach and how Gen Z repurposed the Skull.

Emoji fascinate me — full disclosure, I submitted proposals that helped bring the Saluting Face, Bubbles, Waffle, and Yawning Face emoji into the world — so I’m to share more of the emoji universe with Installer readers.

Here’s Keith’s homescreen and his explanation of what’s on it.

Image: Keith Broni

The phone: Google Pixel 6. Like Molly last week, I’m an “until the wheels fall off” type of person when it comes to my personal phone, though I use an iPhone 15 Pro for work.

The wallpaper: A 2019 picture of me (I’m in all black, bar the white soles of my sneakers) wandering the grounds of the Zōjō-ji Temple in Tokyo, with the Tokyo Tower looming in the background.

  • The top row is my fitness-focused stuff, as I’m preparing for a Hyrox-like competition in October, so it’s all step counts (Garmin Connect), protein goals (MyFitnessPal), habit tracking (HabitShare — a free and simple tracker with some nice friend-sharing elements for accountability), and gym class bookings (Glofox).
  • Since I’m European, WhatsApp is my primary mode of texting / calling folks. And since I’m Irish, you’ve got my Irish banking app, AIB Mobile, ha.
  • I’m a millennial, so Instagram is my primary mode of social media these days, though I’m also on X / Bluesky / TikTok a lot for work purposes (that end up drifting into non-work purposes if I’m not careful).
  • I’m a Letterboxd user, and do enjoy attending the cinema as much as I can (shoutout to the Lighthouse in Dublin — I was there last night watching Friendship, which I thoroughly enjoyed). But really, YouTube is where I spend far too much of my time (and to a lesser extent, Nebula, which is on the next page of my home screen).

I also asked Keith to share a few things he’s into right now. Here’s what he said:

  • Music-wise, I’ve been listening to a lot of Oasis on Spotify lately. I was one of the lucky 2 percent who managed to snag tickets for their current run of shows, and I’ve been riding the high ever since their homecoming gig in Manchester on Sunday (the 20th). I’ve also been enjoying several tracks from the new Wet Leg album — a band that famously workshopped their name using various emoji combos — as well as two new songs from The Last Dinner Party and Nine Inch Nails.
  • To tie in with the fitness app bit above, I’ve been playing the Final Fantasy Pixel Remasters while walking on a home treadmill. I’ve been a big old-school JRPG fan since I was a kid, though I’d only played IV from this collection before.
  • This Instagram account @theblack.lodge, who’s been doing limited drops of T-shirts inspired by David Lynch films (I’m a big fan, especially of Twin Peaks, RIP). I’ve bought at least one item from every drop so far.
  • Book-wise, I’ve been listening to Algospeak: How Social Media Is Transforming the Future of Language by Adam Aleksic on Audible (sitting on the second page of my home screen). Before this, I finished reading the physical copy of Face with Tears of Joy: A Natural History of Emoji by Keith Houston, in which our work at Emojipedia is heavily referenced (and I’m honored to have been directly name-dropped in).
  • Another bit of a self-serving one (especially since you mentioned it yourself last week), but I’ve been enjoying dipping into the revived Emojitracker.com site during spare moments to watch live emoji usage unfold.

Here’s what the Installer community is into this week. I want to know what you’re into right now as well! Email installer@theverge.com with your recommendations for anything and everything, and we’ll feature some of our favorites here every week. For even more great recommendations, check out the replies to this post on The Verge, this post on Threads, and this post on Bluesky.

“I use Outplay on my Apple Watch to track the soccer matches I play. It shows a heat map of my movement during a match and has metrics specific to playing soccer like sprints. It works better for me than using the built-in Apple Workout app or Strava.” – Harry Tequila

“Been playing the macOS version of Cyberpunk 2077 on a base M4 Mac Mini. I am impressed with the game and graphic quality of this version. Just hoping this is the start for more Mac-native versions of A & S tier games.” – penguinchiller

“I just completed the (not so) Endless Tower in Clair Obscur: Expedition 33 and beat Clea… on an MSI Claw 8 with an Intel chip! Didn’t expect that after panning the first Claw. It’s the only handheld I’ve really enjoyed this game on so far!” – Sean

“I’ve become obsessed with the Twos app and using it constantly. Highly recommend!” – Rich

“After steadfastly avoiding it because of its difficulty, I recently purchased and have been dying over and over again in Returnal. The environment really does feel like a hostile alien planet, and with headphones it feels like I can hear every drop of rain. And yeah, it’s really difficult.” – Aram

“Been playing Lushfoil Photography Sim and it is so relaxing and lovely.” – Allison

“I’ve been binging CityNerd on YouTube. The channel discusses cities, transportation, urban design, walkability, etc. Good stuff.” – low_light_mixes

Last week, I included a tidbit from a reader who has been playing Dance Dance Revolution recently. As someone who played a lot of DDR when I was a kid (“Butterfly,” anyone?), I was curious to learn more about how he does it. Here’s some of what the reader, Tom, told me — he also pointed me to a starter guide on Reddit:

“I’ll start with acknowledging I am fully not an expert in the scene, but spent a lot of time in the local arcade and at home playing on a PS2 when I was growing up.

“I grabbed the Deluxe 1-inch thick foam pad from www.ddrpad.com, along with the extension cord and the preloaded Stepmania Flash drive. Stepmania is the open-source clone of DDR. People make their own stepcharts (the arrows that scroll up during a song) and make custom songs that you can download and load in, as well as songs from the actual DDR games.

“Someone who is really into playing may want to eventually invest in a metal pad, like what’s in the arcades. They’re super sturdy and will last for years of use, whereas soft mats could potentially tear over time. There are many manufacturers out there, and the price range is usually from $400 and up for the metal pads. People even buy old arcade cabinets for thousands of dollars and have them in their home.”

Thanks, Tom! See you next week, everyone!

Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.




Source link
#Apple #beta #season

Previously, DoorDash worked with Alphabet’s drone delivery division Wing on a number of pilots in places like Dallas-Fort Worth, Virginia, and Australia. Now the company says it will deploy its own drones, designed specifically for food delivery. The drones will be designed and assembled within the company’s DoorDash Labs division, which houses its robotics and AI teams responsible for building its Dot sidewalk delivery robots.

“We’re designing an aircraft purpose-built for the gaps we see in local commerce, and it complements the short- and long-range designs our partners are scaling,” the company said in a press release. “Our drone is also proudly American-designed and built, with the majority of components made right here in the U.S.” (To be sure, the vast majority of drones are manufactured in China.)

In addition, DoorDash says it will also build all the necessary ground-level infrastructure to enable restaurants to hand off their deliveries to drones. As seen with other drone delivery pilots, there is a complex system of structures and additional equipment that needs to be built on the ground to enable drones to pickup their payloads before embarking on their deliveries. And there are flight plans, airspace management, and drop-off logistics that needs to be established to track the drones in flight. DoorDash says its committed to building all that infrastructure itself.

“The aircraft is what everyone sees,” the company says. “The harder problem is the infrastructure and integration on the ground, and that’s our advantage. Solving hard physical-world problems for local businesses is our focus, from real-time inventory reconciliation to universal handoff systems for drive-throughs, rooftops, or merchant back doors. We are building that complete end-to-end system.”

DoorDash says drone delivery could solve the problem of midrange deliveries that still end up taking longer because couriers are hard to find. The company says more than 20 percent of its orders involve distances of three to five miles, but that those orders can take on average nearly 25 percent longer than shorter deliveries, largely because finding a courier takes longer for these mid-range trips.

“Routing mid-range deliveries to drones lets Dashers focus on the orders they often favor: shorter deliveries that can be done quickly, staying near a high concentration of merchants for optimal routing, and maximizing their earning potential by getting them to their next order faster,” the company says.

DoorDash didn’t reveal which markets it was targeting for its initial launch, noting that it would announce more details in the months to come.

#DoorDash #airborne #drone #delivery #divisionDrones,Food,News,Science,Tech,Transportation">DoorDash is going airborne with new drone delivery divisionDoorDash is launching a new drone delivery program called DoorDash Air. The largest food delivery app in the US said that it has approval from the Federal Aviation Administration that clears the way for drone delivery in the near future.DoorDash said it has received a Part 135 air carrier certification from the FAA that enable it to delivery small packages via drone. Other companies to receive this designation include Alphabet’s Wing, Zipline, and Amazon.Previously, DoorDash worked with Alphabet’s drone delivery division Wing on a number of pilots in places like Dallas-Fort Worth, Virginia, and Australia. Now the company says it will deploy its own drones, designed specifically for food delivery. The drones will be designed and assembled within the company’s DoorDash Labs division, which houses its robotics and AI teams responsible for building its Dot sidewalk delivery robots.“We’re designing an aircraft purpose-built for the gaps we see in local commerce, and it complements the short- and long-range designs our partners are scaling,” the company said in a press release. “Our drone is also proudly American-designed and built, with the majority of components made right here in the U.S.” (To be sure, the vast majority of drones are manufactured in China.)In addition, DoorDash says it will also build all the necessary ground-level infrastructure to enable restaurants to hand off their deliveries to drones. As seen with other drone delivery pilots, there is a complex system of structures and additional equipment that needs to be built on the ground to enable drones to pickup their payloads before embarking on their deliveries. And there are flight plans, airspace management, and drop-off logistics that needs to be established to track the drones in flight. DoorDash says its committed to building all that infrastructure itself.“The aircraft is what everyone sees,” the company says. “The harder problem is the infrastructure and integration on the ground, and that’s our advantage. Solving hard physical-world problems for local businesses is our focus, from real-time inventory reconciliation to universal handoff systems for drive-throughs, rooftops, or merchant back doors. We are building that complete end-to-end system.”DoorDash says drone delivery could solve the problem of midrange deliveries that still end up taking longer because couriers are hard to find. The company says more than 20 percent of its orders involve distances of three to five miles, but that those orders can take on average nearly 25 percent longer than shorter deliveries, largely because finding a courier takes longer for these mid-range trips.“Routing mid-range deliveries to drones lets Dashers focus on the orders they often favor: shorter deliveries that can be done quickly, staying near a high concentration of merchants for optimal routing, and maximizing their earning potential by getting them to their next order faster,” the company says.DoorDash didn’t reveal which markets it was targeting for its initial launch, noting that it would announce more details in the months to come.#DoorDash #airborne #drone #delivery #divisionDrones,Food,News,Science,Tech,Transportation

Previously, DoorDash worked with Alphabet’s drone delivery division Wing on a number of pilots in places like Dallas-Fort Worth, Virginia, and Australia. Now the company says it will deploy its own drones, designed specifically for food delivery. The drones will be designed and assembled within the company’s DoorDash Labs division, which houses its robotics and AI teams responsible for building its Dot sidewalk delivery robots.

“We’re designing an aircraft purpose-built for the gaps we see in local commerce, and it complements the short- and long-range designs our partners are scaling,” the company said in a press release. “Our drone is also proudly American-designed and built, with the majority of components made right here in the U.S.” (To be sure, the vast majority of drones are manufactured in China.)

In addition, DoorDash says it will also build all the necessary ground-level infrastructure to enable restaurants to hand off their deliveries to drones. As seen with other drone delivery pilots, there is a complex system of structures and additional equipment that needs to be built on the ground to enable drones to pickup their payloads before embarking on their deliveries. And there are flight plans, airspace management, and drop-off logistics that needs to be established to track the drones in flight. DoorDash says its committed to building all that infrastructure itself.

“The aircraft is what everyone sees,” the company says. “The harder problem is the infrastructure and integration on the ground, and that’s our advantage. Solving hard physical-world problems for local businesses is our focus, from real-time inventory reconciliation to universal handoff systems for drive-throughs, rooftops, or merchant back doors. We are building that complete end-to-end system.”

DoorDash says drone delivery could solve the problem of midrange deliveries that still end up taking longer because couriers are hard to find. The company says more than 20 percent of its orders involve distances of three to five miles, but that those orders can take on average nearly 25 percent longer than shorter deliveries, largely because finding a courier takes longer for these mid-range trips.

“Routing mid-range deliveries to drones lets Dashers focus on the orders they often favor: shorter deliveries that can be done quickly, staying near a high concentration of merchants for optimal routing, and maximizing their earning potential by getting them to their next order faster,” the company says.

DoorDash didn’t reveal which markets it was targeting for its initial launch, noting that it would announce more details in the months to come.

#DoorDash #airborne #drone #delivery #divisionDrones,Food,News,Science,Tech,Transportation">DoorDash is going airborne with new drone delivery division

DoorDash is launching a new drone delivery program called DoorDash Air. The largest food delivery app in the US said that it has approval from the Federal Aviation Administration that clears the way for drone delivery in the near future.

DoorDash said it has received a Part 135 air carrier certification from the FAA that enable it to delivery small packages via drone. Other companies to receive this designation include Alphabet’s Wing, Zipline, and Amazon.

Previously, DoorDash worked with Alphabet’s drone delivery division Wing on a number of pilots in places like Dallas-Fort Worth, Virginia, and Australia. Now the company says it will deploy its own drones, designed specifically for food delivery. The drones will be designed and assembled within the company’s DoorDash Labs division, which houses its robotics and AI teams responsible for building its Dot sidewalk delivery robots.

“We’re designing an aircraft purpose-built for the gaps we see in local commerce, and it complements the short- and long-range designs our partners are scaling,” the company said in a press release. “Our drone is also proudly American-designed and built, with the majority of components made right here in the U.S.” (To be sure, the vast majority of drones are manufactured in China.)

In addition, DoorDash says it will also build all the necessary ground-level infrastructure to enable restaurants to hand off their deliveries to drones. As seen with other drone delivery pilots, there is a complex system of structures and additional equipment that needs to be built on the ground to enable drones to pickup their payloads before embarking on their deliveries. And there are flight plans, airspace management, and drop-off logistics that needs to be established to track the drones in flight. DoorDash says its committed to building all that infrastructure itself.

“The aircraft is what everyone sees,” the company says. “The harder problem is the infrastructure and integration on the ground, and that’s our advantage. Solving hard physical-world problems for local businesses is our focus, from real-time inventory reconciliation to universal handoff systems for drive-throughs, rooftops, or merchant back doors. We are building that complete end-to-end system.”

DoorDash says drone delivery could solve the problem of midrange deliveries that still end up taking longer because couriers are hard to find. The company says more than 20 percent of its orders involve distances of three to five miles, but that those orders can take on average nearly 25 percent longer than shorter deliveries, largely because finding a courier takes longer for these mid-range trips.

“Routing mid-range deliveries to drones lets Dashers focus on the orders they often favor: shorter deliveries that can be done quickly, staying near a high concentration of merchants for optimal routing, and maximizing their earning potential by getting them to their next order faster,” the company says.

DoorDash didn’t reveal which markets it was targeting for its initial launch, noting that it would announce more details in the months to come.

#DoorDash #airborne #drone #delivery #divisionDrones,Food,News,Science,Tech,Transportation
New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised $9 million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 

Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 

For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

As AI content floods the internet, Pangram raises M to detect it | TechCrunch
New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised  million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 







Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 


For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.







The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a -per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.







My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.#content #floods #internet #Pangram #raises #detect #TechCrunchai slop
Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.

The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a $20-per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.

My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

#content #floods #internet #Pangram #raises #detect #TechCrunchai slop">As AI content floods the internet, Pangram raises M to detect it | TechCrunch
New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised  million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 







Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 


For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.







The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a -per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.







My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.#content #floods #internet #Pangram #raises #detect #TechCrunchai slop

Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised $9 million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 

Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 

For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

As AI content floods the internet, Pangram raises M to detect it | TechCrunch
New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised  million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 







Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 


For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.







The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a -per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.







My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.#content #floods #internet #Pangram #raises #detect #TechCrunchai slop
Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.

The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a $20-per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.

My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

#content #floods #internet #Pangram #raises #detect #TechCrunchai slop">As AI content floods the internet, Pangram raises $9M to detect it | TechCrunch

New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised $9 million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 

Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 

For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

As AI content floods the internet, Pangram raises M to detect it | TechCrunch
New York-based AI detection startup Pangram is on a mission to combat the AI slop infestation spreading across the internet, and it just raised  million on a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow. 

Pangram’s fundraise — led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza — comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image. 







Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, plus it can more easily detect AI humanizer programs. The AI image detector is only available via research preview for now; Pangram plans to release it more widely in the coming weeks.

Stanford AI and machine learning grads Max Spero and Bradley Emi launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an internet full of bots, AI-generated SEO slop content, and what Spero calls “LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter.”

“I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not,” Spero told TechCrunch. “Especially text that you’re reading, because it changes how people approach the text. Is this something that I’m going to have to look out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?”

Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length, and tone of voice, but written by a frontier LLM. 

“Our model is learning the stylistic differences and the choices that AI makes consistently and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector isn’t relying on copy-paste metadata or hidden watermarks. 


For Pangram, AI detection isn’t just about whether or not a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance — like in the case of someone who writes something themselves, but then asks AI to edit or clean it up. Spero believes AI assistance can be acceptable, just so long as the writer discloses their use of AI. 

Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.







The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a -per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.







My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.#content #floods #internet #Pangram #raises #detect #TechCrunchai slop
Image Credits:Pangram

Pangram’s emergence comes at a time when AI usage is becoming more commonplace. In some cases, like the Canadian politician who read an AI prompt aloud in a speech to lawmakers, the mistakes result in ridicule. In other cases, as with certain lawyers making their case using fake citations created by ChatGPT, the consequences could be sanctions and fines.  

That backlash isn’t just costing individuals embarrassment or sanctions — it’s starting to show up in institutional rules, too.

The open-access archive arXiv introduced a new enforcement policy this year, stating that submissions containing evidence that authors failed to review LLM output (like hallucinated references or meta comments such as, “Would you like me to make any changes?”) can trigger a one-year submission ban.

Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks, and GPTZero are are chasing the same demand, each building its own detector.

Pangram’s technology, while not perfect, could help fuel the resistance to accepting the AI-generated content flooding the internet, the courtroom, and academic papers. 

Users can access Pangram via a $20-per-month subscription on the web or download the Chrome extension, which automatically labels posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen. 

Pangram also offers its technology via API. Notably, Substack recently integrated Pangram’s technology into its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers and agents, and recruiters, among others, per Spero. 

Does Pangram work?

Pangram detected AI-generated content even when lightly edited by a human. Image Credits:Pangram/TechCrunch

Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. At the same time, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content.  

I also gave ChatGPT and Claude one of my own articles and asked them to polish it up. Pangram gave it a 13% AI assisted score, which was probably close to accurate, but the model was able to detect subtle word-choice changes in some sentences and ignored them in others. It also flagged some sentences as AI-assisted when they were human written. That was notable because when I gave Pangram that same article in its entirety, as I had written it, it got a 100% human score. 

Maybe the problem was that news articles can be a bit dry and could easily sound like AI. So I tried a different tactic. I tested Pangram on my own more voicey, personal Substack newsletter content, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.

My limited testing of Pangram’s new image detection model turned out to be equally impressive. 

Image Credits:Pangram/TechCrunch

Pangram’s AI image detection system promises to spot AI-generated images across AI models, unlike OpenAI’s or Google DeepMind’s watermark-based checks, which mostly detect their own output. It works on pixel-level distributions, learning subtle statistical differences between real photos and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photo. 

In my testing, the model easily detected AI-generated imagery, whether it was photorealistic or cartoonish. I can also confirm the model could detect an AI image appearing in a real-world photo — the heat map Pangram provides clearly lighting up over the image — though in one instance it incorrectly labeled a photo of an AI-generated image as human content. 

Spero says he doesn’t want his technology to fuel a witch hunt against people using AI for writing, but that there needs to be some sort of mechanism to push back against the slop. 

“The future that I see is that AI content just continues to proliferate,” Spero said. “We’re getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we’re just gonna see more and more AI, and it’s just gonna drown out any human signal that we have.”

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

#content #floods #internet #Pangram #raises #detect #TechCrunchai slop

Post Comment