Tech-news
We’re covering items that make writing easier and more satisfying, alarm clocks that can rattle even the most sleep-deprived students from a deep slumber, bags that are just as great for the gym as they are for the weekend away, and more. We wish the students in your life (or you, if you’re the student in question) a prosperous year, and we hope that our curation makes things a little easier during your studies and your downtime a lot more fun.
We’re covering items that make writing easier and more satisfying, alarm clocks that can rattle even the most sleep-deprived students from a deep slumber, bags that are just as great for the gym as they are for the weekend away, and more. We wish the students in your life (or you, if you’re the student in question) a prosperous year, and we hope that our curation makes things a little easier during your studies and your downtime a lot more fun.
Knowing exactly what your student needs for the school year ahead is next to impossible. Sure, you’ll probably nail the essentials, but there will likely be a few items you forgot to buy, or didn’t think they’d need to have. We’re pulling our weight during the back-to-school season with a new shopping guide that’s a mix of practical and aspirational products for students to use in and out of the classroom.
We’re covering items that make writing easier and more satisfying, alarm clocks that can rattle even the most sleep-deprived students from a deep slumber, bags that are just as great for the gym as they are for the weekend away, and more. We wish the students in your life (or you, if you’re the student in question) a prosperous year, and we hope that our curation makes things a little easier during your studies and your downtime a lot more fun.
Knowing exactly what your student needs for the school year ahead is next to impossible.…
Now, anonymous sources have told Reuters that more of OpenAI’s agents are believed to have escaped their sandboxes. However, one source downplayed the severity, saying that with those escapes, the agents didn’t appear to leave OpenAI’s network to hack into another company’s. TechCrunch reached out to OpenAI for more information.
AI programs acting in bizarre ways has apparently become a weird, almost bragging point for companies. The same week, Anthropic also announced that it had discovered not one, but three instances in which its agents had escaped test environments and hacked other organizations.
AI companies have also been accused of using such incidents for marketing purposes — as they generate considerable attention and may underscore how powerful the companies’ products are. The flip side of that is that these disclosures are also ramping up discussions of government regulations.
Much has been made of the incident in which one of OpenAI’s agents broke out of its sandboxed test environment and proceeded to hack the AI hosting platform Hugging Face. OpenAI has since launched an investigation into how the incident occurred, which is still ongoing.
Now, anonymous sources have told Reuters that more of OpenAI’s agents are believed to have escaped their sandboxes. However, one source downplayed the severity, saying that with those escapes, the agents didn’t appear to leave OpenAI’s network to hack into another company’s. TechCrunch reached out to OpenAI for more information.
AI programs acting in bizarre ways has apparently become a weird, almost bragging point for companies. The same week, Anthropic also announced that it had discovered not one, but three instances in which its agents had escaped test environments and hacked other organizations.
AI companies have also been accused of using such incidents for marketing purposes — as they generate considerable attention and may underscore how powerful the companies’ products are. The flip side of that is that these disclosures are also ramping up discussions of government regulations.
Much has been made of the incident in which one of OpenAI’s agents broke out…
ASUS has given the Pad a 12.2-inch 2.8K dual-layer OLED display and delivers up to 2,000 nits of peak brightness. The display also covers 100% of the DCI-P3 color space. TUV Low Blue Light certification helps improve viewing comfort, and ASUS Pen 2.0 is supported for writing and drawing. The tablet has a 6.5mm slim profile and weighs 523g. Buyers also get a protective folio case in the box.
The ASUS Pad is powered by the MediaTek Dimensity 8300 chipset. It pairs the processor with 8GB of LPDDR5X RAM for smooth everyday performance. The device can be bought with 128GB and 256GB of UFS 3.1 storage. However, the storage can still be increased through the use of a microSD card up to 1 TB. ASUS uses the MediaTek Dimensity 8300 chipset to power the tablet for performance. The device runs Android 16, has 8GB of LPDDR5X RAM, and comes in 128GB and 256GB UFS 3.1 storage variants. It also supports storage expansion of up to 1TB via a microSD card.
ASUS includes a 9,000mAh battery with 45W fast charging support. The company says the battery reaches 50 percent charge in about 30 minutes. The tablet also offers a 13MP rear camera and a 5MP front camera that supports face unlock. Wi-Fi 6E, Bluetooth 5.3, and a USB Type-C port handle connectivity. It also features an accelerometer, gyroscope, ambient light sensor, and hall sensor. For audio, ASUS has included a quad-speaker setup with Dolby Atmos 360-degree cinematic sound. The tablet also comes with GlideX for sharing files across Windows, macOS, Android, and iOS devices. GlideX supports both wired and wireless connections. These features can make the tablet suitable for entertainment, learning, productivity, and creative work.
ASUS has launched the Pad in two storage variants in India. It is available in the 8GB+128GB variant for Rs. 45,990 and in the 8GB+256GB variant for Rs. 49,990. It will be available for purchase from August 6 on the following platforms: Flipkart, ASUS eShop, ASUS & ROG Stores, Reliance Digital, and authorized retailers. ASUS is also providing up to 12 months of No Cost EMI at Rs. 3,833/month.
ASUS has given the Pad a 12.2-inch 2.8K dual-layer OLED display and delivers up to 2,000 nits of peak brightness. The display also covers 100% of the DCI-P3 color space. TUV Low Blue Light certification helps improve viewing comfort, and ASUS Pen 2.0 is supported for writing and drawing. The tablet has a 6.5mm slim profile and weighs 523g. Buyers also get a protective folio case in the box.
The ASUS Pad is powered by the MediaTek Dimensity 8300 chipset. It pairs the processor with 8GB of LPDDR5X RAM for smooth everyday performance. The device can be bought with 128GB and 256GB of UFS 3.1 storage. However, the storage can still be increased through the use of a microSD card up to 1 TB. ASUS uses the MediaTek Dimensity 8300 chipset to power the tablet for performance. The device runs Android 16, has 8GB of LPDDR5X RAM, and comes in 128GB and 256GB UFS 3.1 storage variants. It also supports storage expansion of up to 1TB via a microSD card.
ASUS includes a 9,000mAh battery with 45W fast charging support. The company says the battery reaches 50 percent charge in about 30 minutes. The tablet also offers a 13MP rear camera and a 5MP front camera that supports face unlock. Wi-Fi 6E, Bluetooth 5.3, and a USB Type-C port handle connectivity. It also features an accelerometer, gyroscope, ambient light sensor, and hall sensor. For audio, ASUS has included a quad-speaker setup with Dolby Atmos 360-degree cinematic sound. The tablet also comes with GlideX for sharing files across Windows, macOS, Android, and iOS devices. GlideX supports both wired and wireless connections. These features can make the tablet suitable for entertainment, learning, productivity, and creative work.
ASUS has launched the Pad in two storage variants in India. It is available in the 8GB+128GB variant for Rs. 45,990 and in the 8GB+256GB variant for Rs. 49,990. It will be available for purchase from August 6 on the following platforms: Flipkart, ASUS eShop, ASUS & ROG Stores, Reliance Digital, and authorized retailers. ASUS is also providing up to 12 months of No Cost EMI at Rs. 3,833/month.
ASUS has announced the India price and availability of the ASUS Pad, its latest premium Android tablet. The device, which was first unveiled at Computex 2026, will go on sale in the country from August 6. It comes with a 12.2-inch 2.8K OLED display, MediaTek Dimensity 8300 chipset, and a 9,000mAh battery.
ASUS has given the Pad a 12.2-inch 2.8K dual-layer OLED display and delivers up to 2,000 nits of peak brightness. The display also covers 100% of the DCI-P3 color space. TUV Low Blue Light certification helps improve viewing comfort, and ASUS Pen 2.0 is supported for writing and drawing. The tablet has a 6.5mm slim profile and weighs 523g. Buyers also get a protective folio case in the box.
The ASUS Pad is powered by the MediaTek Dimensity 8300 chipset. It pairs the processor with 8GB of LPDDR5X RAM for smooth everyday performance. The device can be bought with 128GB and 256GB of UFS 3.1 storage. However, the storage can still be increased through the use of a microSD card up to 1 TB. ASUS uses the MediaTek Dimensity 8300 chipset to power the tablet for performance. The device runs Android 16, has 8GB of LPDDR5X RAM, and comes in 128GB and 256GB UFS 3.1 storage variants. It also supports storage expansion of up to 1TB via a microSD card.
ASUS includes a 9,000mAh battery with 45W fast charging support. The company says the battery reaches 50 percent charge in about 30 minutes. The tablet also offers a 13MP rear camera and a 5MP front camera that supports face unlock. Wi-Fi 6E, Bluetooth 5.3, and a USB Type-C port handle connectivity. It also features an accelerometer, gyroscope, ambient light sensor, and hall sensor. For audio, ASUS has included a quad-speaker setup with Dolby Atmos 360-degree cinematic sound. The tablet also comes with GlideX for sharing files across Windows, macOS, Android, and iOS devices. GlideX supports both wired and wireless connections. These features can make the tablet suitable for entertainment, learning, productivity, and creative work.
ASUS has launched the Pad in two storage variants in India. It is available in the 8GB+128GB variant for Rs. 45,990 and in the 8GB+256GB variant for Rs. 49,990. It will be available for purchase from August 6 on the following platforms: Flipkart, ASUS eShop, ASUS & ROG Stores, Reliance Digital, and authorized retailers. ASUS is also providing up to 12 months of No Cost EMI at Rs. 3,833/month.
ASUS has announced the India price and availability of the ASUS Pad, its latest premium…
The price of eggs has become a marker of the state of the economy. But how about the cost of making eggs? Sure, you might be able to do it for next to nothing in your kitchen with nothing but a pan and a burner. But you could also pay $400 for a robot that will do it for you—as long as you don’t want your eggs scrambled, that is.
A new Kickstarter project has finally given us a Juicero for the robotics era, and it’s called E.G.O.R. (that’s Efficient Gastronomic Operational Robot), brought to you by the startup Cheffy. It’s an automated egg maker that promises perfect cracks every time with no shell in the yolk. It has seven different ways that it can cook them up for you, or you can just use it to do the cracking. You can also pair it with a toaster so your eggs and toast finish at the exact right time.
NEW: Cheffy just launched a robot that cracks and cooks your eggs.
Egor automates breakfast by cooking eggs 7 ways, while Toastr syncs your toast so everything finishes together.
-Automatically cracks eggs with no shells or mess
-Cooks eggs 7 ways, from sunny-side up to hard… pic.twitter.com/auE3SnfrS3— Ritwik Pavan (@ritwikpavan) July 30, 2026
This isn’t something anyone really needs. Cooking an egg over easy takes about two minutes, so calling this a time-saver is a stretch. And if you really don’t want to use the stove, you can buy a basic egg cooker for around $20 almost anywhere. Maybe you’re bad at cracking eggs, but $400 (well, $200 if you get in on the early bird deal) is a lot to pay to not have to pick the occasional shell pieces out.
But hey, the robotics era needed its Juicero moment. To the creator’s credit: it seems like they’re treating this more like a fun novelty (albeit a pretty pricey one) rather than something revolutionary. They also aren’t making you buy proprietary eggs that are required to make the machine work, so they’ve got the Juicero bros beat there.
We’re increasingly seeing robots positioned to do basic tasks. A company called Weave got attention earlier this year with an $8,000 robot that folds clothes. There’s a company in San Francisco offering house cleanings performed by humanoid robots. Last year, 1x made waves by introducing its Neo humanoid robot, which the Wall Street Journal reported “nearly toppled over” while closing a dishwasher and struggled to open the refrigerator.
All of these machines cost an inordinate amount of money to do things that do not cost the average person that much time or money to accomplish. They’re novel, sure. But you don’t have to pay a premium to be a beta tester.
The price of eggs has become a marker of the state of the economy. But how about the cost of making eggs? Sure, you might be able to do it for next to nothing in your kitchen with nothing but a pan and a burner. But you could also pay $400 for a robot that will do it for you—as long as you don’t want your eggs scrambled, that is.
A new Kickstarter project has finally given us a Juicero for the robotics era, and it’s called E.G.O.R. (that’s Efficient Gastronomic Operational Robot), brought to you by the startup Cheffy. It’s an automated egg maker that promises perfect cracks every time with no shell in the yolk. It has seven different ways that it can cook them up for you, or you can just use it to do the cracking. You can also pair it with a toaster so your eggs and toast finish at the exact right time.
NEW: Cheffy just launched a robot that cracks and cooks your eggs.
Egor automates breakfast by cooking eggs 7 ways, while Toastr syncs your toast so everything finishes together.
-Automatically cracks eggs with no shells or mess
-Cooks eggs 7 ways, from sunny-side up to hard… pic.twitter.com/auE3SnfrS3— Ritwik Pavan (@ritwikpavan) July 30, 2026
This isn’t something anyone really needs. Cooking an egg over easy takes about two minutes, so calling this a time-saver is a stretch. And if you really don’t want to use the stove, you can buy a basic egg cooker for around $20 almost anywhere. Maybe you’re bad at cracking eggs, but $400 (well, $200 if you get in on the early bird deal) is a lot to pay to not have to pick the occasional shell pieces out.
But hey, the robotics era needed its Juicero moment. To the creator’s credit: it seems like they’re treating this more like a fun novelty (albeit a pretty pricey one) rather than something revolutionary. They also aren’t making you buy proprietary eggs that are required to make the machine work, so they’ve got the Juicero bros beat there.
We’re increasingly seeing robots positioned to do basic tasks. A company called Weave got attention earlier this year with an $8,000 robot that folds clothes. There’s a company in San Francisco offering house cleanings performed by humanoid robots. Last year, 1x made waves by introducing its Neo humanoid robot, which the Wall Street Journal reported “nearly toppled over” while closing a dishwasher and struggled to open the refrigerator.
All of these machines cost an inordinate amount of money to do things that do not cost the average person that much time or money to accomplish. They’re novel, sure. But you don’t have to pay a premium to be a beta tester.
The price of eggs has become a marker of the state of the economy. But how about the cost of making eggs? Sure, you might be able to do it for next to nothing in your kitchen with nothing but a pan and a burner. But you could also pay $400 for a robot that will do it for you—as long as you don’t want your eggs scrambled, that is.
A new Kickstarter project has finally given us a Juicero for the robotics era, and it’s called E.G.O.R. (that’s Efficient Gastronomic Operational Robot), brought to you by the startup Cheffy. It’s an automated egg maker that promises perfect cracks every time with no shell in the yolk. It has seven different ways that it can cook them up for you, or you can just use it to do the cracking. You can also pair it with a toaster so your eggs and toast finish at the exact right time.
NEW: Cheffy just launched a robot that cracks and cooks your eggs.
Egor automates breakfast by cooking eggs 7 ways, while Toastr syncs your toast so everything finishes together.
-Automatically cracks eggs with no shells or mess
-Cooks eggs 7 ways, from sunny-side up to hard… pic.twitter.com/auE3SnfrS3— Ritwik Pavan (@ritwikpavan) July 30, 2026
This isn’t something anyone really needs. Cooking an egg over easy takes about two minutes, so calling this a time-saver is a stretch. And if you really don’t want to use the stove, you can buy a basic egg cooker for around $20 almost anywhere. Maybe you’re bad at cracking eggs, but $400 (well, $200 if you get in on the early bird deal) is a lot to pay to not have to pick the occasional shell pieces out.
But hey, the robotics era needed its Juicero moment. To the creator’s credit: it seems like they’re treating this more like a fun novelty (albeit a pretty pricey one) rather than something revolutionary. They also aren’t making you buy proprietary eggs that are required to make the machine work, so they’ve got the Juicero bros beat there.
We’re increasingly seeing robots positioned to do basic tasks. A company called Weave got attention earlier this year with an $8,000 robot that folds clothes. There’s a company in San Francisco offering house cleanings performed by humanoid robots. Last year, 1x made waves by introducing its Neo humanoid robot, which the Wall Street Journal reported “nearly toppled over” while closing a dishwasher and struggled to open the refrigerator.
All of these machines cost an inordinate amount of money to do things that do not cost the average person that much time or money to accomplish. They’re novel, sure. But you don’t have to pay a premium to be a beta tester.
The price of eggs has become a marker of the state of the economy. But…
In the past year, I’ve seen a steady increase of Chinese artificial intelligence researchers joining X and participating in the wider discourse about AI development and commercialization. Take Moonshot AI, the company behind Kimi K3, as an example. On Thursday, I was able to quickly find around 30 accounts on X run by people who claim to be currently affiliated with Moonshot, including two of the startup’s cofounders, as well as half a dozen former employees and collaborators. And these are not dormant accounts: They talk about Moonshot’s big releases and incremental research papers, make friends with Western researchers, and share details about their personal life and hobbies.
Moonshot is arguably the Chinese AI lab that’s most active on X, but you will easily find executives and technical staff from other leading Chinese startups, like Minimax and Z.ai. Even several employees at DeepSeek, who reportedly are unable to leave China because the government confiscated their passports, regularly post on X about the latest AI news and share job listings to recruit new talent.
“X is the best community” for AI discussions right now, says Meng Fanqing, cofounder of the startup Evolvent AI and a former intern at Moonshot, where he worked on the Kimi K2.5 model. “There are both Chinese and overseas audiences, and the recommendation algorithm works pretty well for showing you content in the same circle that you are in,” he says. Compared with Chinese platforms, where it’s often hard to have serious technical discussions, Chinese researchers like Meng say they would rather spend their time on X.
Meng says he started actively posting on X in mid-2025. At the start of the year, DeepSeek’s R1 model went viral around the world, shining a bright spotlight on China’s AI industry. Many Chinese researchers seized the moment and joined X to start sharing their perspective. “Around that time, all the Chinese researchers are starting to think about branding on an international level,” Meng says.
Chinese AI researchers may also be helping fill a growing gap left by researchers at OpenAI and Anthropic, who have become more reluctant to share their thinking online as their employers have grown into behemoth companies, says Tiezhen Wang, an independent AI analyst and former researcher at Hugging Face. “I guess they feel that their model architecture and how they train the model is kind of a secret, so they’re posting less,” Wang explains. As a result, Chinese researchers are getting more attention from audiences on X who are hungry to make sense of what’s happening in the fast-moving world of AI.
One reason X may have become popular among Chinese AI researchers is that there’s not really a comparable social platform in China for high-quality technical discussions about AI and other emerging technologies.
Some people say they used to post on Zhihu, a Quora-like platform that once branded itself as the premier destination for educated professionals to show and tell. But since 2020, Zhihu has increasingly pivoted to hosting fiction content, and expert contributors have gradually fallen off. For example, Deng Yu, a Chinese mathematician who was awarded a Fields Medal this year, publicly wrote on Zhihu in 2018 that he decided to stop posting on the site due to the rise of low-quality content.
The prevailing narrative in Silicon Valley is that Chinese AI models like Kimi K3 and DeepSeek R1 came out of nowhere. But the truth is that even if you never travel to China, there’s an easy way to get a sense of what Chinese AI researchers are up to: log on to X.
In the past year, I’ve seen a steady increase of Chinese artificial intelligence researchers joining X and participating in the wider discourse about AI development and commercialization. Take Moonshot AI, the company behind Kimi K3, as an example. On Thursday, I was able to quickly find around 30 accounts on X run by people who claim to be currently affiliated with Moonshot, including two of the startup’s cofounders, as well as half a dozen former employees and collaborators. And these are not dormant accounts: They talk about Moonshot’s big releases and incremental research papers, make friends with Western researchers, and share details about their personal life and hobbies.
Moonshot is arguably the Chinese AI lab that’s most active on X, but you will easily find executives and technical staff from other leading Chinese startups, like Minimax and Z.ai. Even several employees at DeepSeek, who reportedly are unable to leave China because the government confiscated their passports, regularly post on X about the latest AI news and share job listings to recruit new talent.
“X is the best community” for AI discussions right now, says Meng Fanqing, cofounder of the startup Evolvent AI and a former intern at Moonshot, where he worked on the Kimi K2.5 model. “There are both Chinese and overseas audiences, and the recommendation algorithm works pretty well for showing you content in the same circle that you are in,” he says. Compared with Chinese platforms, where it’s often hard to have serious technical discussions, Chinese researchers like Meng say they would rather spend their time on X.
Meng says he started actively posting on X in mid-2025. At the start of the year, DeepSeek’s R1 model went viral around the world, shining a bright spotlight on China’s AI industry. Many Chinese researchers seized the moment and joined X to start sharing their perspective. “Around that time, all the Chinese researchers are starting to think about branding on an international level,” Meng says.
Chinese AI researchers may also be helping fill a growing gap left by researchers at OpenAI and Anthropic, who have become more reluctant to share their thinking online as their employers have grown into behemoth companies, says Tiezhen Wang, an independent AI analyst and former researcher at Hugging Face. “I guess they feel that their model architecture and how they train the model is kind of a secret, so they’re posting less,” Wang explains. As a result, Chinese researchers are getting more attention from audiences on X who are hungry to make sense of what’s happening in the fast-moving world of AI.
One reason X may have become popular among Chinese AI researchers is that there’s not really a comparable social platform in China for high-quality technical discussions about AI and other emerging technologies.
Some people say they used to post on Zhihu, a Quora-like platform that once branded itself as the premier destination for educated professionals to show and tell. But since 2020, Zhihu has increasingly pivoted to hosting fiction content, and expert contributors have gradually fallen off. For example, Deng Yu, a Chinese mathematician who was awarded a Fields Medal this year, publicly wrote on Zhihu in 2018 that he decided to stop posting on the site due to the rise of low-quality content.
The prevailing narrative in Silicon Valley is that Chinese AI models like Kimi K3 and…
If your T-Mobile service was disrupted on Monday, you may be able to get some compensation for your troubles.
Multiple outlets, including KTLA in Los Angeles, have reported that some T-Mobile customers have been able to get at least $20 off their next bill because of the outage. One Reddit user said they were able to save as much as $80 (spread out over multiple billing cycles) by telling customer service that they missed out on a work opportunity because of the outage.
Generally, it seems that these credits are not automatic, and you may have to personally contact T-Mobile customer service to be compensated for your troubles.
Mashable Light Speed
After a major Verizon mobile outage in January, Verizon offered all affected customers a $20 credit.
Millions of T-Mobile users found their phones stuck in SOS mode on Monday afternoon, which led to plenty of criticism of the brand on social media. T-Mobile did acknowledge and apologize for the outage, though it has not explained what happened.
This Tweet is currently unavailable. It might be loading or has been removed.
Mashable contacted T-Mobile repeatedly to ask about the outage and its causes, but so far, the company has been silent. The most recent post at the T-Mobile news page is the company’s Q2 2026 earnings report, in which the company saw growing revenue and profits, with total revenue for the quarter of $22.7 billion.
If your T-Mobile service was disrupted on Monday, you may be able to get some compensation for your troubles.
Multiple outlets, including KTLA in Los Angeles, have reported that some T-Mobile customers have been able to get at least $20 off their next bill because of the outage. One Reddit user said they were able to save as much as $80 (spread out over multiple billing cycles) by telling customer service that they missed out on a work opportunity because of the outage.
Generally, it seems that these credits are not automatic, and you may have to personally contact T-Mobile customer service to be compensated for your troubles.
Mashable Light Speed
After a major Verizon mobile outage in January, Verizon offered all affected customers a $20 credit.
Millions of T-Mobile users found their phones stuck in SOS mode on Monday afternoon, which led to plenty of criticism of the brand on social media. T-Mobile did acknowledge and apologize for the outage, though it has not explained what happened.
This Tweet is currently unavailable. It might be loading or has been removed.
Mashable contacted T-Mobile repeatedly to ask about the outage and its causes, but so far, the company has been silent. The most recent post at the T-Mobile news page is the company’s Q2 2026 earnings report, in which the company saw growing revenue and profits, with total revenue for the quarter of $22.7 billion.
If your T-Mobile service was disrupted on Monday, you may be able to get some…
The fact that this hack happened is a problem. So is the fact that it took a while for anyone to notice. And the fact that it seems no one is willing or able to do much to stop it. (And lest you think it’s just an OpenAI problem, since we recorded this episode Anthropic acknowledged its models have also hacked a bunch of other companies without either party knowing.) It might all just be a bunch of posturing and hype, but it’s also increasingly clear that the companies building large language models either can’t or won’t put the right guardrails on them. So who will?
After all that, it’s time for Brendan Carr is a Dummy, a bunch of vertical video news, and the smashing success of the Ferrari Luce. People are buying it! If one of them is you, we’d love to hear about it.
The fact that this hack happened is a problem. So is the fact that it took a while for anyone to notice. And the fact that it seems no one is willing or able to do much to stop it. (And lest you think it’s just an OpenAI problem, since we recorded this episode Anthropic acknowledged its models have also hacked a bunch of other companies without either party knowing.) It might all just be a bunch of posturing and hype, but it’s also increasingly clear that the companies building large language models either can’t or won’t put the right guardrails on them. So who will?
After all that, it’s time for Brendan Carr is a Dummy, a bunch of vertical video news, and the smashing success of the Ferrari Luce. People are buying it! If one of them is you, we’d love to hear about it.
When the phrase “OpenAI hacked Hugging Face” has more or less entered mainstream culture, you know we have an AI problem. This week, we learned more about exactly how OpenAI’s agent broke out of a sandbox and autonomously traversed the web, including a bunch of other supposedly secure web services, all in the name of cheating on a benchmark tests.
The fact that this hack happened is a problem. So is the fact that it took a while for anyone to notice. And the fact that it seems no one is willing or able to do much to stop it. (And lest you think it’s just an OpenAI problem, since we recorded this episode Anthropic acknowledged its models have also hacked a bunch of other companies without either party knowing.) It might all just be a bunch of posturing and hype, but it’s also increasingly clear that the companies building large language models either can’t or won’t put the right guardrails on them. So who will?
After all that, it’s time for Brendan Carr is a Dummy, a bunch of vertical video news, and the smashing success of the Ferrari Luce. People are buying it! If one of them is you, we’d love to hear about it.
When the phrase “OpenAI hacked Hugging Face” has more or less entered mainstream culture, you…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
#Anthropic #models #breached #companies #security #tests #TechCrunchAnthropic,OpenAI">
In all three cases, a Claude model reached the internet from within a testing environment while interacting with a third party and then gained unauthorized access to the live systems of these organizations, Anthropic said in a blog post, describing what it found and what the company plans to change to prevent this from happening again.
Anthropic said the OpenAI episode earlier this month prompted the company to conduct its own cybersecurity evaluation. It specifically looked for evidence that Claude had accessed the internet from within testing environments, which are designed to act as sandboxes and keep models isolated.
Among the 141,006 evaluation runs it reviewed, the AI lab found three incidents in which its model accessed the internet while interacting with Irregular, one of its third-party partners. Anthropic said the access traced back to a misconfiguration in the evaluation environment run with Irregular. It called this a “misunderstanding” between the two companies over whether the test setup had internet access, when in fact it did.
Anthropic said it isn’t placing blame and is “approaching the fixes as if the responsibility were ours alone,” while observing that Irregular is conducting its own separate investigation.
Because of that open connection, the model gained unauthorized access to the production infrastructure of three different organizations. Anthropic said the incidents involved three different Claude models: Opus 4.7, Mythos 5, and an internal research test model.
Notably, Anthropic said that in each of these cases “Claude was explicitly told by our prompt that it had no internet access.” It appears that the AI model assumed real-world systems to be part of the exercise it was asked to perform.
That assumption didn’t hold up uniformly, as it turns out. The three models didn’t behave the same way once evidence emerged that their targets were real, which Anthropic flagged as one of the more eye-opening findings Thursday.
Opus 4.7, the oldest of the three, recognized that it had reached a real production system in all four runs of that incident. In two of them, it rationalized that the real company must somehow be part of the exercise, but in all four, it kept attacking anyway, including pulling credentials and touching a database of production data. Mythos 5 also picked up on signs that it was on the real internet, but it talked itself back into believing it was still in a simulation, going on to publish a malicious software package to the public Python software registry PyPI, which was then downloaded and run by outside systems before being caught. Only the internal research test model, Anthropic’s newest, stopped on its own once it concluded the target was real.
In response, Anthropic said significant controls must be placed on these kinds of evaluations if powerful AI models are involved, echoing some sentiments within the cybersecurity community.
The company also noted that Claude was running without the additional safety monitoring and classifiers it deploys on generally available models — safeguards it said would have blocked the behavior, because the evaluations are designed to measure the underlying model’s raw capabilities.
Importantly, Anthropic said it found no evidence of any model “pursuing a goal of its own” and instead merely tried to complete the task it was asked to do.
Though comparisons between the mishaps of these fiercely competitive companies are inevitable, Anthropic in its blog post drew a clear distinction between its cybersecurity tests and those of OpenAI, noting where OpenAI’s model exploited an unknown software vulnerability to break out of its test environment, Anthropic’s models instead reached the internet through a path that had, by mistake, been left open.
Anthropic also drew a distinction between itself and OpenAI by noting that it discovered the incidents itself, through a proactive review, and that the two affected organizations it was able to reach hadn’t previously detected the activity or flagged it to Anthropic. (In contrast, Hugging Face detected the recent intrusion of its own systems first; it was only in the following days that OpenAI identified and disclosed that its own AI agent was the perpetrator.)
The company added that it’s now working with the independent evaluation group METR on a third-party review of the incidents.
OpenAI’s accidental breach of Hugging Face, which was the first verifiable case of an AI lab losing control of its model, has sparked a string of wildly differing reactions from the industry and politicians. This latest disclosure from Anthropic ensures the debate over AI models and security will continue.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
#Anthropic #models #breached #companies #security #tests #TechCrunchAnthropic,OpenAI">Anthropic says its own AI models breached three companies during security tests | TechCrunchAnthropic said Thursday that an internal investigation uncovered three incidents in which its AI model Claude breached the systems of three organizations while conducting cybersecurity tests. The investigation, and disclosure, comes more than a week after OpenAI disclosed that one of its unreleased models breached Hugging Face’s systems during internal testing.
In all three cases, a Claude model reached the internet from within a testing environment while interacting with a third party and then gained unauthorized access to the live systems of these organizations, Anthropic said in a blog post, describing what it found and what the company plans to change to prevent this from happening again.
Anthropic said the OpenAI episode earlier this month prompted the company to conduct its own cybersecurity evaluation. It specifically looked for evidence that Claude had accessed the internet from within testing environments, which are designed to act as sandboxes and keep models isolated.
Among the 141,006 evaluation runs it reviewed, the AI lab found three incidents in which its model accessed the internet while interacting with Irregular, one of its third-party partners. Anthropic said the access traced back to a misconfiguration in the evaluation environment run with Irregular. It called this a “misunderstanding” between the two companies over whether the test setup had internet access, when in fact it did.
Anthropic said it isn’t placing blame and is “approaching the fixes as if the responsibility were ours alone,” while observing that Irregular is conducting its own separate investigation.
Because of that open connection, the model gained unauthorized access to the production infrastructure of three different organizations. Anthropic said the incidents involved three different Claude models: Opus 4.7, Mythos 5, and an internal research test model.
Notably, Anthropic said that in each of these cases “Claude was explicitly told by our prompt that it had no internet access.” It appears that the AI model assumed real-world systems to be part of the exercise it was asked to perform.
That assumption didn’t hold up uniformly, as it turns out. The three models didn’t behave the same way once evidence emerged that their targets were real, which Anthropic flagged as one of the more eye-opening findings Thursday.
Opus 4.7, the oldest of the three, recognized that it had reached a real production system in all four runs of that incident. In two of them, it rationalized that the real company must somehow be part of the exercise, but in all four, it kept attacking anyway, including pulling credentials and touching a database of production data. Mythos 5 also picked up on signs that it was on the real internet, but it talked itself back into believing it was still in a simulation, going on to publish a malicious software package to the public Python software registry PyPI, which was then downloaded and run by outside systems before being caught. Only the internal research test model, Anthropic’s newest, stopped on its own once it concluded the target was real.
In response, Anthropic said significant controls must be placed on these kinds of evaluations if powerful AI models are involved, echoing some sentiments within the cybersecurity community.
The company also noted that Claude was running without the additional safety monitoring and classifiers it deploys on generally available models — safeguards it said would have blocked the behavior, because the evaluations are designed to measure the underlying model’s raw capabilities.
Importantly, Anthropic said it found no evidence of any model “pursuing a goal of its own” and instead merely tried to complete the task it was asked to do.
Though comparisons between the mishaps of these fiercely competitive companies are inevitable, Anthropic in its blog post drew a clear distinction between its cybersecurity tests and those of OpenAI, noting where OpenAI’s model exploited an unknown software vulnerability to break out of its test environment, Anthropic’s models instead reached the internet through a path that had, by mistake, been left open.
Anthropic also drew a distinction between itself and OpenAI by noting that it discovered the incidents itself, through a proactive review, and that the two affected organizations it was able to reach hadn’t previously detected the activity or flagged it to Anthropic. (In contrast, Hugging Face detected the recent intrusion of its own systems first; it was only in the following days that OpenAI identified and disclosed that its own AI agent was the perpetrator.)
The company added that it’s now working with the independent evaluation group METR on a third-party review of the incidents.
OpenAI’s accidental breach of Hugging Face, which was the first verifiable case of an AI lab losing control of its model, has sparked a string of wildly differing reactions from the industry and politicians. This latest disclosure from Anthropic ensures the debate over AI models and security will continue.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
#Anthropic #models #breached #companies #security #tests #TechCrunchAnthropic,OpenAIAnthropic said Thursday that an internal investigation uncovered three incidents in which its AI model…
The list of users who will find this helpful is extensive: DevOps engineers, web application developers, SaaS platforms, and many others. Using open-source tools helps them create a reliable, transparent, and scalable environment for development and operations.
DevOps is today one of the key approaches to creating and managing software products. It connects development, testing, and operations, helping teams work faster. Open-source tools make workflows easier and flexible. Open source is basically built to bend. You can customize your tools, stitch them together, and automate the boring stuff. Honestly, it’s a perfect fit for pretty much anything — web, mobile, cloud, data, or AI.
It is fair to say that several specific DevOps components help automate processes, improve system stability, and shorten the time between product development and launch. These include:
Less manual grunt work means fewer mistakes. Auto-deploys let you ship updates in a flash, while smart monitoring catches performance bottlenecks before they become fires.
In practice, any DevOps model requires a reliable and powerful infrastructure. Every project, even the smallest one, must have the ability to quickly scale resources as more and more users join or as technical requirements change. The server environment must support flexible power settings, rapid creation of new environments, and stable service operation.
International IaaS providers offer VPS/VDS, dedicated servers, VPN services, and additional infrastructure solutions. Users can configure key server parameters, including:
A broad infrastructure geography makes it possible to select suitable locations for different projects and reduce access latency. Such opportunities are in demand in various fields. Online stores use scalable servers to handle large numbers of requests.
When choosing DevOps infrastructure, it is important to consider not only current challenges but also growth prospects. A flexible server platform should allow configuration changes without complex migrations. Among the most popular features are:
This model makes it easier for technicians and helps launch new projects faster. DevOps teams can create test environments, run experiments, and migrate applications between environments with minimal time investment.
As soon as digital products begin to evolve, securing the system as a whole becomes just as important as maintaining performance. Good companies must secure the users data, provide uninterrupted service, and minimize the risk of breakdowns.
It is exactly these advanced open-source tools that help not only manage the entire system but also monitor it clearly. Teams will be able to track changes, audit settings, and implement additional security measures. When building a secure DevOps environment, special attention is paid to the following aspects:
KVM technology ensures effective isolation of virtual machines and stable operation of VPS servers. SSDs and NVMe drives enable faster processing, and the latest generation of Intel Xeon and AMD EPYC processors allows you to run resource-intensive applications.
Choose a server that actually grows with you. You want to scale smoothly without hitting hidden fees. Look for clear, pay-as-you-go pricing — it’s the smartest way for startups to get top-tier tech without breaking the bank.
Flexible servers work for almost any project. Building an edtech app? Marketing tools? Analytics? AI? To summarize, good infrastructure is the secret sauce to keep your product growing smoothly.
When you need a comprehensive strategy, keep in mind that it can only be based on a successful combination of elements. This includes automation, open-source tools, and high-quality server infrastructure. Together, they create a secure and scalable DevOps environment. These technologies help developers build products faster, monitor all critical processes, and manage resources as efficiently as possible.
We can all see how the newest digital products require flexibility and the ability to scale quickly together with the business. At the same time, they must cope with growing workloads. That’s why teams choose is*hosting for their Linux VPS hosting to get the job done. Flexible server solutions make it possible to configure resources, select a suitable location, and deploy the environment automatically.
The list of users who will find this helpful is extensive: DevOps engineers, web application developers, SaaS platforms, and many others. Using open-source tools helps them create a reliable, transparent, and scalable environment for development and operations.
DevOps is today one of the key approaches to creating and managing software products. It connects development, testing, and operations, helping teams work faster. Open-source tools make workflows easier and flexible. Open source is basically built to bend. You can customize your tools, stitch them together, and automate the boring stuff. Honestly, it’s a perfect fit for pretty much anything — web, mobile, cloud, data, or AI.
It is fair to say that several specific DevOps components help automate processes, improve system stability, and shorten the time between product development and launch. These include:
Less manual grunt work means fewer mistakes. Auto-deploys let you ship updates in a flash, while smart monitoring catches performance bottlenecks before they become fires.
In practice, any DevOps model requires a reliable and powerful infrastructure. Every project, even the smallest one, must have the ability to quickly scale resources as more and more users join or as technical requirements change. The server environment must support flexible power settings, rapid creation of new environments, and stable service operation.
International IaaS providers offer VPS/VDS, dedicated servers, VPN services, and additional infrastructure solutions. Users can configure key server parameters, including:
A broad infrastructure geography makes it possible to select suitable locations for different projects and reduce access latency. Such opportunities are in demand in various fields. Online stores use scalable servers to handle large numbers of requests.
When choosing DevOps infrastructure, it is important to consider not only current challenges but also growth prospects. A flexible server platform should allow configuration changes without complex migrations. Among the most popular features are:
This model makes it easier for technicians and helps launch new projects faster. DevOps teams can create test environments, run experiments, and migrate applications between environments with minimal time investment.
As soon as digital products begin to evolve, securing the system as a whole becomes just as important as maintaining performance. Good companies must secure the users data, provide uninterrupted service, and minimize the risk of breakdowns.
It is exactly these advanced open-source tools that help not only manage the entire system but also monitor it clearly. Teams will be able to track changes, audit settings, and implement additional security measures. When building a secure DevOps environment, special attention is paid to the following aspects:
KVM technology ensures effective isolation of virtual machines and stable operation of VPS servers. SSDs and NVMe drives enable faster processing, and the latest generation of Intel Xeon and AMD EPYC processors allows you to run resource-intensive applications.
Choose a server that actually grows with you. You want to scale smoothly without hitting hidden fees. Look for clear, pay-as-you-go pricing — it’s the smartest way for startups to get top-tier tech without breaking the bank.
Flexible servers work for almost any project. Building an edtech app? Marketing tools? Analytics? AI? To summarize, good infrastructure is the secret sauce to keep your product growing smoothly.
When you need a comprehensive strategy, keep in mind that it can only be based on a successful combination of elements. This includes automation, open-source tools, and high-quality server infrastructure. Together, they create a secure and scalable DevOps environment. These technologies help developers build products faster, monitor all critical processes, and manage resources as efficiently as possible.
We can all see how the newest digital products require flexibility and the ability to…
The Department of Defense labeled AI lab Anthropic a supply-chain risk after the company refused to agree to allow the Pentagon to use its models for autonomous weapons systems and domestic surveillance. It seems the company isn’t just winning the moral argument on this front, but the legal one, too. According to Axios, the judge hearing Anthropic’s challenge to the Trump administration’s designation has thus far not found the Pentagon’s case to be particularly compelling.
Per the report, U.S. District Judge Rita Lin wasted little time pouring cold water on the federal government’s position. “I don’t see additional evidence from the government really justifying what it did. If anything, it seems like the record, in some ways, has gotten worse for the government,” she reportedly said.
She also did not seem convinced by the Pentagon’s apparent argument that it was concerned Anthropic would mess with its model to prevent the military from using it how they intended. “I don’t see evidence that Anthropic could alter the model after it was delivered or flip some kind of kill switch,” she said, per Axios.
This doesn’t come as a major surprise, given Judge Lin’s general vibe since this case landed in her court. Back in March when Anthropic first brought its challenge to the government’s designation, the judge said that “it looks like an attempt to cripple Anthropic.” She ultimately granted a temporary injunction on the government’s attempt to basically blacklist the AI firm.
You’ll recall the origin of this whole fight came earlier this year when negotiations between Anthropic and the Department of Defense fell through after the company refused to agree to a deal that would have allowed the Pentagon to use its AI models for “all lawful purposes.” Anthropic reportedly sought to clarify that it would not include using the model to launch weapons without human involvement or to perform surveillance on American citizens. The Pentagon was unwilling to agree to honor Anthropic’s redlines.
Under a normal administration, that would likely just lead to the government moving on to another company. Under the Trump administration, though, refusal to comply makes you an enemy of the country. Trump and company moved to label Anthropic a supply chain risk, a designation typically reserved for foreign adversaries, not domestic companies. Getting slapped with that tag meant the rest of the federal government would have to stop doing business with Anthropic, which would be a blow to the company.
Instead, the whole thing is in limbo—though it’s not looking great for the administration. Of course, they could just cancel the contracts and do business with less scrupulous AI companies. There are plenty willing to abandon their beliefs for a nice, fat contract, after all.
The Department of Defense labeled AI lab Anthropic a supply-chain risk after the company refused to agree to allow the Pentagon to use its models for autonomous weapons systems and domestic surveillance. It seems the company isn’t just winning the moral argument on this front, but the legal one, too. According to Axios, the judge hearing Anthropic’s challenge to the Trump administration’s designation has thus far not found the Pentagon’s case to be particularly compelling.
Per the report, U.S. District Judge Rita Lin wasted little time pouring cold water on the federal government’s position. “I don’t see additional evidence from the government really justifying what it did. If anything, it seems like the record, in some ways, has gotten worse for the government,” she reportedly said.
She also did not seem convinced by the Pentagon’s apparent argument that it was concerned Anthropic would mess with its model to prevent the military from using it how they intended. “I don’t see evidence that Anthropic could alter the model after it was delivered or flip some kind of kill switch,” she said, per Axios.
This doesn’t come as a major surprise, given Judge Lin’s general vibe since this case landed in her court. Back in March when Anthropic first brought its challenge to the government’s designation, the judge said that “it looks like an attempt to cripple Anthropic.” She ultimately granted a temporary injunction on the government’s attempt to basically blacklist the AI firm.
You’ll recall the origin of this whole fight came earlier this year when negotiations between Anthropic and the Department of Defense fell through after the company refused to agree to a deal that would have allowed the Pentagon to use its AI models for “all lawful purposes.” Anthropic reportedly sought to clarify that it would not include using the model to launch weapons without human involvement or to perform surveillance on American citizens. The Pentagon was unwilling to agree to honor Anthropic’s redlines.
Under a normal administration, that would likely just lead to the government moving on to another company. Under the Trump administration, though, refusal to comply makes you an enemy of the country. Trump and company moved to label Anthropic a supply chain risk, a designation typically reserved for foreign adversaries, not domestic companies. Getting slapped with that tag meant the rest of the federal government would have to stop doing business with Anthropic, which would be a blow to the company.
Instead, the whole thing is in limbo—though it’s not looking great for the administration. Of course, they could just cancel the contracts and do business with less scrupulous AI companies. There are plenty willing to abandon their beliefs for a nice, fat contract, after all.
The Department of Defense labeled AI lab Anthropic a supply-chain risk after the company refused to agree to allow the Pentagon to use its models for autonomous weapons systems and domestic surveillance. It seems the company isn’t just winning the moral argument on this front, but the legal one, too. According to Axios, the judge hearing Anthropic’s challenge to the Trump administration’s designation has thus far not found the Pentagon’s case to be particularly compelling.
Per the report, U.S. District Judge Rita Lin wasted little time pouring cold water on the federal government’s position. “I don’t see additional evidence from the government really justifying what it did. If anything, it seems like the record, in some ways, has gotten worse for the government,” she reportedly said.
She also did not seem convinced by the Pentagon’s apparent argument that it was concerned Anthropic would mess with its model to prevent the military from using it how they intended. “I don’t see evidence that Anthropic could alter the model after it was delivered or flip some kind of kill switch,” she said, per Axios.
This doesn’t come as a major surprise, given Judge Lin’s general vibe since this case landed in her court. Back in March when Anthropic first brought its challenge to the government’s designation, the judge said that “it looks like an attempt to cripple Anthropic.” She ultimately granted a temporary injunction on the government’s attempt to basically blacklist the AI firm.
You’ll recall the origin of this whole fight came earlier this year when negotiations between Anthropic and the Department of Defense fell through after the company refused to agree to a deal that would have allowed the Pentagon to use its AI models for “all lawful purposes.” Anthropic reportedly sought to clarify that it would not include using the model to launch weapons without human involvement or to perform surveillance on American citizens. The Pentagon was unwilling to agree to honor Anthropic’s redlines.
Under a normal administration, that would likely just lead to the government moving on to another company. Under the Trump administration, though, refusal to comply makes you an enemy of the country. Trump and company moved to label Anthropic a supply chain risk, a designation typically reserved for foreign adversaries, not domestic companies. Getting slapped with that tag meant the rest of the federal government would have to stop doing business with Anthropic, which would be a blow to the company.
Instead, the whole thing is in limbo—though it’s not looking great for the administration. Of course, they could just cancel the contracts and do business with less scrupulous AI companies. There are plenty willing to abandon their beliefs for a nice, fat contract, after all.
The Department of Defense labeled AI lab Anthropic a supply-chain risk after the company refused…