×
Election night at Kalshi HQ

Election night at Kalshi HQ

This is an excerpt of Sources by Alex Heath, a newsletter about AI and the tech industry, syndicated just for The Verge subscribers once a week.

At 8PM on election night in New York City, I arrived at an unmarked office building in the Meatpacking District.

Inside, a few dozen young Kalshi employees moved between clusters of desks, pizza boxes, and a large projector displaying live markets for the day’s key races. The vibe was quiet but focused. On the screen, numbers flickered as bets adjusted in real time.

Near the projector, co-founders Tarek Mansour and Luana Lopes Lara chatted with a CBS News crew filming a segment for the next morning. CBS had just called the Virginia governor’s race. Mansour pointed out that Kalshi’s market had predicted the result almost an hour earlier.

“We’re doing a billion dollars in transaction volume a week now.”

I expected a trading floor atmosphere. Instead, the office felt subdued. “I think it’s quieter than usual because there’s less volatility on this one,” Mansour told me later from a small conference room. The New York mayor’s race had long been priced as a landslide. Zohran Mamdani had held a roughly 95 percent chance of winning on Kalshi (and its rival Polymarket) even before polls closed. Still, about $100 million in trades on the New York race went through Kalshi that day.

In recent months, I’ve been tracking the rise of prediction markets and particularly Kalshi. Despite being federally licensed and much larger than Polymarket, it’s the latter that dominates the conversation in tech circles. Mansour wants to change that.

Kalshi’s betting page for the New York City mayoral election, captured one day after the election.

“Kalshi is arguably one of — maybe the — fastest-growing companies in America this year,” he told me. “We’re doing a billion dollars in transaction volume a week now.” Last year, the company saw just $300 million for the entire year. Mansour declined to share revenue figures, but even at a 1–2 percent fee per trade, the math suggests that business is booming.

Three factors have fueled that growth this year: securing a federal license to operate, expanding into sports betting, and striking a partnership with Robinhood to power prediction markets. While sports have been a major draw, Mansour’s ambitions go far beyond that.

“I think prediction markets are the next generation of the stock market,” he said. “They have media consequences. Everyone is an expert on something — everyone has opinions. These markets give those opinions a price.”

Kalshi called the New Jersey governor’s race 32 minutes before any news outlet

He hinted at new partnerships with media outlets and even entertainment event tie-ins. “We’re doing a lot with news networks in the coming months,” he said. “If the truth that comes out of these markets becomes mainstream, we’ve basically achieved our mission.”

Given how new prediction markets are, Kalshi and Polymarket still need to prove that they can remain reliable sources for predicting elections. Fox News took a reputational hit for accidentally calling Arizona for Joe Biden too early in 2020. Meanwhile, Kalshi and Polymarket brag about calling races even before results are in. If one of them gets a key race wrong, it could call into question the legitimacy of prediction markets.

With less than an hour left before polls closed, Mansour showed me Kalshi data from the New York mayoral race. Voters in the city were buying Andrew Cuomo contracts more heavily, but Mamdani dominated elsewhere. He was winning among women and younger traders; Cuomo’s support skewed older and male.

As we spoke, Kalshi called the New Jersey governor’s race at 8:20PM — 32 minutes before any news outlet. Mansour compared Kalshi’s role to that of financial markets: “Should the stock market replace bank analysts? No. Analysts provide input, and the market finds the real price. We’re doing the same thing for events.”

I asked whether people constantly text him for predictions, especially on an election night. He laughed. “Yeah. But I tell them: just look at the market. I don’t have any extra information.”

As 9PM neared, I assumed he’d stay in the office as polls closed. But as I stepped into my Uber, I saw him dart out and get into another car down the street.

He didn’t need to wait. Kalshi called the New York race for Mamdani one minute after polls closed and 36 minutes before any media outlet.

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


Source link
#Election #night #Kalshi

Snapchat is officially joining the race to clean AI slop out of social media.

The company announced July 31 that wholly AI-generated videos will no longer be eligible for recommendation in Spotlight, Snapchat’s discovery feed for trending videos from creators across the platform in order to “reward authentic creativity.”

The policy does not prevent users from posting AI-generated videos elsewhere on Snapchat. It also does not disqualify every video that involves AI. Creators can continue using Snapchat’s AI tools to edit or enhance their original footage, and those videos will remain eligible for Spotlight recommendations with transparency indicators.

The line Snapchat is drawing comes down to who, or what, created the underlying video. AI can help a person make something. It cannot make the entire thing and still expect Snapchat’s recommendation system to distribute it.

“As low-quality, repetitive, AI-generated content becomes increasingly common across the internet, we want Spotlight to remain a place where people can discover authentic creativity from real people,” the company wrote in a blog post.

Keeping entirely generated videos out of Spotlight matters on the platform because recommendation is what allow a creator’s work to reach people who do not already follow them. Removing that distribution makes Spotlight far less valuable as a growth channel for automated content farms that can produce synthetic videos at scale.

It also protects a part of Snapchat that has been attracting more creators. Snap said the number of unique people contributing to Spotlight globally has increased by more than 120 percent over the past year.

Exactly how Snapchat will identify wholly AI-generated videos is less clear. The company has not explained what level of human involvement would make a video eligible or how it will distinguish an AI-enhanced recording from a generated video that received a few human edits.

Snap acknowledged that challenge in its announcement, saying that “no detection system is perfect.”

Social platforms are reconsidering AI slop

Snapchat isn’t the first platform to decide that the flood of repetitive AI content is becoming a problem for its users.

LinkedIn began expanding tools in July that allow members to report posts and comments that “seem like AI slop.” The company is also introducing classifiers intended to reduce low-quality AI content in recommendations and replacing its “enhance your post” writing feature with a more limited proofreading tool.

LinkedIn Chief Product Officer Hari Srinivasan emphasized that the company does not consider all AI-assisted work to be slop. The concern is content that feels automated, generic, or disconnected from the person publishing it.

Substack has taken a transparency-focused approach. On July 21, the newsletter platform introduced an optional AI-detection feature built with Pangram. Readers can scan posts, notes, replies, and comments longer than 100 words to receive an estimate of how much of the text was written by a person or with AI assistance.

YouTube’s current policies similarly make repetitive or mass-produced “inauthentic content” ineligible for monetization. Its spam rules also prohibit creators from using automated or synthetic tools to flood the platform with large volumes of nearly identical videos.

TikTok has also put more of the decision in viewers’ hands. The platform began testing a setting in November that lets users increase or decrease the amount of AI-generated content that appears in their For You feeds.

Snapchat’s decision also arrives as Meta imagines a very different future for social media.

In its latest earnings remarks on July 29, Meta said its Muse image and video models would create a “nearly infinite universe of personalized content” across its platforms. That AI-generated material would become another source of recommendations alongside posts from friends and established creators.

Snapchat is making a different bet for Spotlight. As synthetic video becomes easier to produce and harder to distinguish from recorded footage, the platform is reserving its most valuable discovery space for content that begins with a person.

Drawing that line may be easier than enforcing it. Still, Snapchat has made its preference clear: AI can stay behind the scenes. Spotlight belongs to the humans.

#Snapchat #bans #fully #AIgenerated #videos #Spotlight #recommendations">Snapchat bans fully AI-generated videos from Spotlight recommendations
                                                            Snapchat is officially joining the race to clean AI slop out of social media.The company announced July 31 that wholly AI-generated videos will no longer be eligible for recommendation in Spotlight, Snapchat’s discovery feed for trending videos from creators across the platform in order to “reward authentic creativity.”The policy does not prevent users from posting AI-generated videos elsewhere on Snapchat. It also does not disqualify every video that involves AI. Creators can continue using Snapchat’s AI tools to edit or enhance their original footage, and those videos will remain eligible for Spotlight recommendations with transparency indicators.

        SEE ALSO:
        
            New safety rules for under-16 Snapchat users
            
        
    
The line Snapchat is drawing comes down to who, or what, created the underlying video. AI can help a person make something. It cannot make the entire thing and still expect Snapchat’s recommendation system to distribute it.“As low-quality, repetitive, AI-generated content becomes increasingly common across the internet, we want Spotlight to remain a place where people can discover authentic creativity from real people,” the company wrote in a blog post.Keeping entirely generated videos out of Spotlight matters on the platform because recommendation is what allow a creator’s work to reach people who do not already follow them. Removing that distribution makes Spotlight far less valuable as a growth channel for automated content farms that can produce synthetic videos at scale.It also protects a part of Snapchat that has been attracting more creators. Snap said the number of unique people contributing to Spotlight globally has increased by more than 120 percent over the past year.Exactly how Snapchat will identify wholly AI-generated videos is less clear. The company has not explained what level of human involvement would make a video eligible or how it will distinguish an AI-enhanced recording from a generated video that received a few human edits.
        
            Mashable Trend Report
        
        
    
Snap acknowledged that challenge in its announcement, saying that “no detection system is perfect.”Social platforms are reconsidering AI slopSnapchat isn’t the first platform to decide that the flood of repetitive AI content is becoming a problem for its users.LinkedIn began expanding tools in July that allow members to report posts and comments that “seem like AI slop.” The company is also introducing classifiers intended to reduce low-quality AI content in recommendations and replacing its “enhance your post” writing feature with a more limited proofreading tool.LinkedIn Chief Product Officer Hari Srinivasan emphasized that the company does not consider all AI-assisted work to be slop. The concern is content that feels automated, generic, or disconnected from the person publishing it.Substack has taken a transparency-focused approach. On July 21, the newsletter platform introduced an optional AI-detection feature built with Pangram. Readers can scan posts, notes, replies, and comments longer than 100 words to receive an estimate of how much of the text was written by a person or with AI assistance.
YouTube’s current policies similarly make repetitive or mass-produced “inauthentic content” ineligible for monetization. Its spam rules also prohibit creators from using automated or synthetic tools to flood the platform with large volumes of nearly identical videos.TikTok has also put more of the decision in viewers’ hands. The platform began testing a setting in November that lets users increase or decrease the amount of AI-generated content that appears in their For You feeds.Snapchat’s decision also arrives as Meta imagines a very different future for social media.In its latest earnings remarks on July 29, Meta said its Muse image and video models would create a “nearly infinite universe of personalized content” across its platforms. That AI-generated material would become another source of recommendations alongside posts from friends and established creators.Snapchat is making a different bet for Spotlight. As synthetic video becomes easier to produce and harder to distinguish from recorded footage, the platform is reserving its most valuable discovery space for content that begins with a person.Drawing that line may be easier than enforcing it. Still, Snapchat has made its preference clear: AI can stay behind the scenes. Spotlight belongs to the humans.

                    
                                            
                            
    
        Topics
                    Artificial Intelligence
                    Snapchat
            

                        
                                    #Snapchat #bans #fully #AIgenerated #videos #Spotlight #recommendations

AI slop out of social media.

The company announced July 31 that wholly AI-generated videos will no longer be eligible for recommendation in Spotlight, Snapchat’s discovery feed for trending videos from creators across the platform in order to “reward authentic creativity.”

The policy does not prevent users from posting AI-generated videos elsewhere on Snapchat. It also does not disqualify every video that involves AI. Creators can continue using Snapchat’s AI tools to edit or enhance their original footage, and those videos will remain eligible for Spotlight recommendations with transparency indicators.

The line Snapchat is drawing comes down to who, or what, created the underlying video. AI can help a person make something. It cannot make the entire thing and still expect Snapchat’s recommendation system to distribute it.

“As low-quality, repetitive, AI-generated content becomes increasingly common across the internet, we want Spotlight to remain a place where people can discover authentic creativity from real people,” the company wrote in a blog post.

Keeping entirely generated videos out of Spotlight matters on the platform because recommendation is what allow a creator’s work to reach people who do not already follow them. Removing that distribution makes Spotlight far less valuable as a growth channel for automated content farms that can produce synthetic videos at scale.

It also protects a part of Snapchat that has been attracting more creators. Snap said the number of unique people contributing to Spotlight globally has increased by more than 120 percent over the past year.

Exactly how Snapchat will identify wholly AI-generated videos is less clear. The company has not explained what level of human involvement would make a video eligible or how it will distinguish an AI-enhanced recording from a generated video that received a few human edits.

Snap acknowledged that challenge in its announcement, saying that “no detection system is perfect.”

Social platforms are reconsidering AI slop

Snapchat isn’t the first platform to decide that the flood of repetitive AI content is becoming a problem for its users.

LinkedIn began expanding tools in July that allow members to report posts and comments that “seem like AI slop.” The company is also introducing classifiers intended to reduce low-quality AI content in recommendations and replacing its “enhance your post” writing feature with a more limited proofreading tool.

LinkedIn Chief Product Officer Hari Srinivasan emphasized that the company does not consider all AI-assisted work to be slop. The concern is content that feels automated, generic, or disconnected from the person publishing it.

Substack has taken a transparency-focused approach. On July 21, the newsletter platform introduced an optional AI-detection feature built with Pangram. Readers can scan posts, notes, replies, and comments longer than 100 words to receive an estimate of how much of the text was written by a person or with AI assistance.

YouTube’s current policies similarly make repetitive or mass-produced “inauthentic content” ineligible for monetization. Its spam rules also prohibit creators from using automated or synthetic tools to flood the platform with large volumes of nearly identical videos.

TikTok has also put more of the decision in viewers’ hands. The platform began testing a setting in November that lets users increase or decrease the amount of AI-generated content that appears in their For You feeds.

Snapchat’s decision also arrives as Meta imagines a very different future for social media.

In its latest earnings remarks on July 29, Meta said its Muse image and video models would create a “nearly infinite universe of personalized content” across its platforms. That AI-generated material would become another source of recommendations alongside posts from friends and established creators.

Snapchat is making a different bet for Spotlight. As synthetic video becomes easier to produce and harder to distinguish from recorded footage, the platform is reserving its most valuable discovery space for content that begins with a person.

Drawing that line may be easier than enforcing it. Still, Snapchat has made its preference clear: AI can stay behind the scenes. Spotlight belongs to the humans.

#Snapchat #bans #fully #AIgenerated #videos #Spotlight #recommendations">Snapchat bans fully AI-generated videos from Spotlight recommendations

Snapchat is officially joining the race to clean AI slop out of social media.

The company announced July 31 that wholly AI-generated videos will no longer be eligible for recommendation in Spotlight, Snapchat’s discovery feed for trending videos from creators across the platform in order to “reward authentic creativity.”

The policy does not prevent users from posting AI-generated videos elsewhere on Snapchat. It also does not disqualify every video that involves AI. Creators can continue using Snapchat’s AI tools to edit or enhance their original footage, and those videos will remain eligible for Spotlight recommendations with transparency indicators.

The line Snapchat is drawing comes down to who, or what, created the underlying video. AI can help a person make something. It cannot make the entire thing and still expect Snapchat’s recommendation system to distribute it.

“As low-quality, repetitive, AI-generated content becomes increasingly common across the internet, we want Spotlight to remain a place where people can discover authentic creativity from real people,” the company wrote in a blog post.

Keeping entirely generated videos out of Spotlight matters on the platform because recommendation is what allow a creator’s work to reach people who do not already follow them. Removing that distribution makes Spotlight far less valuable as a growth channel for automated content farms that can produce synthetic videos at scale.

It also protects a part of Snapchat that has been attracting more creators. Snap said the number of unique people contributing to Spotlight globally has increased by more than 120 percent over the past year.

Exactly how Snapchat will identify wholly AI-generated videos is less clear. The company has not explained what level of human involvement would make a video eligible or how it will distinguish an AI-enhanced recording from a generated video that received a few human edits.

Snap acknowledged that challenge in its announcement, saying that “no detection system is perfect.”

Social platforms are reconsidering AI slop

Snapchat isn’t the first platform to decide that the flood of repetitive AI content is becoming a problem for its users.

LinkedIn began expanding tools in July that allow members to report posts and comments that “seem like AI slop.” The company is also introducing classifiers intended to reduce low-quality AI content in recommendations and replacing its “enhance your post” writing feature with a more limited proofreading tool.

LinkedIn Chief Product Officer Hari Srinivasan emphasized that the company does not consider all AI-assisted work to be slop. The concern is content that feels automated, generic, or disconnected from the person publishing it.

Substack has taken a transparency-focused approach. On July 21, the newsletter platform introduced an optional AI-detection feature built with Pangram. Readers can scan posts, notes, replies, and comments longer than 100 words to receive an estimate of how much of the text was written by a person or with AI assistance.

YouTube’s current policies similarly make repetitive or mass-produced “inauthentic content” ineligible for monetization. Its spam rules also prohibit creators from using automated or synthetic tools to flood the platform with large volumes of nearly identical videos.

TikTok has also put more of the decision in viewers’ hands. The platform began testing a setting in November that lets users increase or decrease the amount of AI-generated content that appears in their For You feeds.

Snapchat’s decision also arrives as Meta imagines a very different future for social media.

In its latest earnings remarks on July 29, Meta said its Muse image and video models would create a “nearly infinite universe of personalized content” across its platforms. That AI-generated material would become another source of recommendations alongside posts from friends and established creators.

Snapchat is making a different bet for Spotlight. As synthetic video becomes easier to produce and harder to distinguish from recorded footage, the platform is reserving its most valuable discovery space for content that begins with a person.

Drawing that line may be easier than enforcing it. Still, Snapchat has made its preference clear: AI can stay behind the scenes. Spotlight belongs to the humans.

#Snapchat #bans #fully #AIgenerated #videos #Spotlight #recommendations

The sale includes the 10-inch Skylight Calendar for $119.99 ($50 off), the 15-inch Skylight Calendar 2 for $259.99 ($20 off), and the 27-inch Skylight Calendar Max for $479.99 ($90 off). The 10-inch model takes up the least amount of space, though the newer, larger 15-inch Calendar 2 features faster performance, a brighter display, and swappable magnetic frames. The 27-inch Calendar Max, meanwhile, offers the sharpest and largest screen and must be mounted to the wall (the other models can stand on a table or be mounted), making it the easiest option for large families to view from across the room.

Regardless of which model you choose, every Skylight Calendar includes the same set of organizational features. Each member of your household can have their own color-coded profile, and each can sync events from Google, Outlook, Apple, Cozi, or Yahoo calendars. You can also create shared lists for groceries, chores, and other to-dos that everyone can access from the calendar or the companion mobile app, and even send your grocery list to Instacart with a couple of taps. Built-in weather forecasts also can help you plan around upcoming events, and you can also share calendar access with family members and trusted contacts.

You get even more perks if you subscribe to the $79 per year Plus Plan (the first month of the service is free to try). Skylight uses AI to automatically turn emails, photos, flyers, and spreadsheets into calendar entries. It also adds meal planning, tools for assigning and rewarding completed chores, and a mode that displays your photos like a digital picture frame.

#Skylights #smart #calendars #backtoschool #saleDeals,Gadgets,Smart Home,Tech,Verge Shopping">Skylight’s smart calendars are up to  off during its back-to-school saleThe start of a new school year can feel hectic, as parents juggle their kids’ classes, extracurriculars and sports on top of work, appointments, and other responsibilities. It’s easy for things to slip through the cracks, but Skylight’s shared smart calendars can help keep everyone on track, and from August 2nd through August 9th they’re up to  off.The sale includes the 10-inch Skylight Calendar for 9.99 ( off), the 15-inch Skylight Calendar 2 for 9.99 ( off), and the 27-inch Skylight Calendar Max for 9.99 ( off). The 10-inch model takes up the least amount of space, though the newer, larger 15-inch Calendar 2 features faster performance, a brighter display, and swappable magnetic frames. The 27-inch Calendar Max, meanwhile, offers the sharpest and largest screen and must be mounted to the wall (the other models can stand on a table or be mounted), making it the easiest option for large families to view from across the room.Regardless of which model you choose, every Skylight Calendar includes the same set of organizational features. Each member of your household can have their own color-coded profile, and each can sync events from Google, Outlook, Apple, Cozi, or Yahoo calendars. You can also create shared lists for groceries, chores, and other to-dos that everyone can access from the calendar or the companion mobile app, and even send your grocery list to Instacart with a couple of taps. Built-in weather forecasts also can help you plan around upcoming events, and you can also share calendar access with family members and trusted contacts.You get even more perks if you subscribe to the  per year Plus Plan (the first month of the service is free to try). Skylight uses AI to automatically turn emails, photos, flyers, and spreadsheets into calendar entries. It also adds meal planning, tools for assigning and rewarding completed chores, and a mode that displays your photos like a digital picture frame.#Skylights #smart #calendars #backtoschool #saleDeals,Gadgets,Smart Home,Tech,Verge Shopping

$119.99 ($50 off), the 15-inch Skylight Calendar 2 for $259.99 ($20 off), and the 27-inch Skylight Calendar Max for $479.99 ($90 off). The 10-inch model takes up the least amount of space, though the newer, larger 15-inch Calendar 2 features faster performance, a brighter display, and swappable magnetic frames. The 27-inch Calendar Max, meanwhile, offers the sharpest and largest screen and must be mounted to the wall (the other models can stand on a table or be mounted), making it the easiest option for large families to view from across the room.

Regardless of which model you choose, every Skylight Calendar includes the same set of organizational features. Each member of your household can have their own color-coded profile, and each can sync events from Google, Outlook, Apple, Cozi, or Yahoo calendars. You can also create shared lists for groceries, chores, and other to-dos that everyone can access from the calendar or the companion mobile app, and even send your grocery list to Instacart with a couple of taps. Built-in weather forecasts also can help you plan around upcoming events, and you can also share calendar access with family members and trusted contacts.

You get even more perks if you subscribe to the $79 per year Plus Plan (the first month of the service is free to try). Skylight uses AI to automatically turn emails, photos, flyers, and spreadsheets into calendar entries. It also adds meal planning, tools for assigning and rewarding completed chores, and a mode that displays your photos like a digital picture frame.

#Skylights #smart #calendars #backtoschool #saleDeals,Gadgets,Smart Home,Tech,Verge Shopping">Skylight’s smart calendars are up to $90 off during its back-to-school sale

The start of a new school year can feel hectic, as parents juggle their kids’ classes, extracurriculars and sports on top of work, appointments, and other responsibilities. It’s easy for things to slip through the cracks, but Skylight’s shared smart calendars can help keep everyone on track, and from August 2nd through August 9th they’re up to $90 off.

The sale includes the 10-inch Skylight Calendar for $119.99 ($50 off), the 15-inch Skylight Calendar 2 for $259.99 ($20 off), and the 27-inch Skylight Calendar Max for $479.99 ($90 off). The 10-inch model takes up the least amount of space, though the newer, larger 15-inch Calendar 2 features faster performance, a brighter display, and swappable magnetic frames. The 27-inch Calendar Max, meanwhile, offers the sharpest and largest screen and must be mounted to the wall (the other models can stand on a table or be mounted), making it the easiest option for large families to view from across the room.

Regardless of which model you choose, every Skylight Calendar includes the same set of organizational features. Each member of your household can have their own color-coded profile, and each can sync events from Google, Outlook, Apple, Cozi, or Yahoo calendars. You can also create shared lists for groceries, chores, and other to-dos that everyone can access from the calendar or the companion mobile app, and even send your grocery list to Instacart with a couple of taps. Built-in weather forecasts also can help you plan around upcoming events, and you can also share calendar access with family members and trusted contacts.

You get even more perks if you subscribe to the $79 per year Plus Plan (the first month of the service is free to try). Skylight uses AI to automatically turn emails, photos, flyers, and spreadsheets into calendar entries. It also adds meal planning, tools for assigning and rewarding completed chores, and a mode that displays your photos like a digital picture frame.

#Skylights #smart #calendars #backtoschool #saleDeals,Gadgets,Smart Home,Tech,Verge Shopping

Post Comment