Thomas Brewster

forbes.comAt OpenAI’s Cyber Lab, Hackers Race To Protect The Internet From AI AttacksUsing the latest GPT model, cyber researchers in OpenAI’s Project Daybreak community found one of the most critical internet bugs in recent memory. It’s part of an all-out sprint to secure the web bef...

🚨NEW🚨 Over the last 2 months, OpenAI's hacker lab coordinated disclosure of *those* PHP bugs. I got an inside peak at that team's work, hearing about the 100s of other wild flaws OpenAI's models are finding daily. Including fun things like Chrome 0-click 0days. www.forbes.com/sites/thomas...

Thomas Brewster reposted

forbes.comAt OpenAI’s Cyber Lab, Hackers Race To Protect The Internet From AI AttacksUsing the latest GPT model, cyber researchers in OpenAI’s Project Daybreak community found one of the most critical internet bugs in recent memory. It’s part of an all-out sprint to secure the web bef...

OpenAI's head of cyber research tells me his team are using AI to find 100 vulns every day. “Many are finding in a week, or in a day, bugs that previously they would have been super excited about finding one a year." Expect repeats of this week's PHP mania soon. www.forbes.com/sites/thomas...

Thomas Brewster reposted

Amy Feldman

@amyfeldman7.bsky.social

forbes.comBillionaires Palmer Luckey And Ben Lamm Spin Up Startup To Genetically Engineer Military DogsBen Lamm of de-extinction startup Colossal has teamed up with Palmer Luckey’s team at Anduril to create a dog that’s optimized for military use. The new “warhounds” will be cloned and gene edited.

*NEW* Colossal Biosciences’ Ben Lamm and Anduril’s Palmer Luckey team up on a new startup to genetically engineer war dogs for the military. Warhound comes out of stealth at $1.2B valuation For @forbes.com with @iainmartin.bsky.social and @mhunterhart.bsky.social www.forbes.com/sites/amyfel...

forbes.comAt OpenAI’s Cyber Lab, Hackers Race To Protect The Internet From AI AttacksUsing the latest GPT model, cyber researchers in OpenAI’s Project Daybreak community found one of the most critical internet bugs in recent memory. It’s part of an all-out sprint to secure the web bef...

OpenAI's head of cyber research tells me his team are using AI to find 100 vulns every day. “Many are finding in a week, or in a day, bugs that previously they would have been super excited about finding one a year." Expect repeats of this week's PHP mania soon. www.forbes.com/sites/thomas...

theatlantic.comThe Man Who Saved the Internet YesterdayOndřej Surý was a relatively anonymous volunteer working on PHP security. Then the internet exploded.

Good interview, but there are more people than this one dude maintaining PHP...

Charlie Warzel@cwarzel.bsky.social

70 percent of the internet was vulnerable yesterday to very bad security flaws inside a coding language. Turns out, *one volunteer* is mostly responsible for maintaining that code & helping push updates. Yesterday Ondřej Surý saved the internet. So I called him up www.theatlantic.com/technology/2...

Thomas Brewster reposted

Monica Hunter-Hart

@mhunterhart.bsky.social

forbes.comBillionaires Palmer Luckey And Ben Lamm Spin Up Startup To Genetically Engineer Military DogsBen Lamm of de-extinction startup Colossal has teamed up with Palmer Luckey’s team at Anduril to create a dog that’s optimized for military use. The new “warhounds” will be cloned and gene edited.

Scoop with @amyfeldman7.bsky.social and @iainmartin.bsky.social - The billionaire trying to resurrect woolly mammoths is now developing bio-enhanced military dogs. Ben Lamm's new startup Warhound raised $200m at a $1.2b valuation and is teaming up with Anduril: www.forbes.com/sites/amyfel...

forbes.comBillionaires Palmer Luckey And Ben Lamm Spin Up Startup To Genetically Engineer Military DogsBen Lamm of de-extinction startup Colossal has teamed up with Palmer Luckey’s team at Anduril to create a dog that’s optimized for military use. The new “warhounds” will be cloned and gene edited.

Wild…

Iain Martin@iainmartin.bsky.social

Colossal Biosciences’ founder has started a new venture to create new bio-enhanced military dogs in collaboration with Anduril’s Palmer Luckey www.forbes.com/sites/amyfel...

The moment we cede reporting to AI in any meaningful way, outside of it being a good tool for analysing large datasets, is the moment we give up entirely on the idea of a free press. Reporting and non-fiction are, like fiction, fundamentally about human communication and human problems.

Thomas Brewster@thomasbrewster.bsky.social

Here's Geoffrey Hinton casually declaring AI will be taking print journalists' jobs pretty soon. It's almost as if he doesn't know what journalism involves. Or understand what it takes for a free press to exist.

Here's Geoffrey Hinton casually declaring AI will be taking print journalists' jobs pretty soon. It's almost as if he doesn't know what journalism involves. Or understand what it takes for a free press to exist.

aws.amazon.comAmazon Threat Intelligence identifies PHP2Shell exploitation prior to public disclosure in MadPot | Amazon Web ServicesAmazon Threat Intelligence identified a PHP2Shell exploitation attempt dating to Oct. 17, 2026, in MadPot telemetry, four days before public disclosure of CVE-2026-109618 and CVE-2026-109619. The expl...

5d

Does anyone have a good explanation for why PHP exploits using the same vulns were seen days ahead of public disclosure and the release of patches? aws.amazon.com/blogs/securi...

forbes.comHackers Hit iPhones Via Hundreds Of Infected WebsitesHundreds of websites with tens of millions of monthly visits are unaware they’re now part of a “delivery network” for iPhone spyware, researchers warn.

5d

🚨NEW🚨 Remember last week's 0-click spyware attacks on PCs via infected ads? Cybercriminals are now doing the same via a hack of the online ad ecosystem. They're targeting iPhones. Just a view of a site or infected ad is enough to get hit. More scary stuff. www.forbes.com/sites/thomas...

forbes.comAI Just Found One Of The Biggest Internet Vulnerabilities EverResearchers claim the AI’s findings “rank among the most severe vulnerabilities discovered in PHP, and arguably in software history.”

6d

🚨 NEW 🚨 AI found those scary PHP bugs because AI is an insane security researcher/hacker now (if that wasn't already painfully obvious). OpenAI and Google models with the exploits this time. www.forbes.com/sites/thomas...

Thomas Brewster reposted

Iain Martin

@iainmartin.bsky.social

1w

TikTok has run dozens of ads promoting the sale of counterfeit currency. One even directed users to a Telegram account offering a discount code for 10% off fake money: “TikTok10.” www.forbes.com/sites/iainma...

forbes.comTikTok Has Cashed In On Ads Pushing Counterfeit CashThe Chinese-owned social network has run dozens of ads promoting the sale of counterfeit currency. One even directed users to a Telegram account offering a discount code for 10% off fake money: “TikTo...

Thomas Brewster reposted

Cynthia Brumfield

@metacurity.com

1w

Whoa. The Federal Trade Commission on Wednesday sent civil subpoenas to AI giants seeking documents about safety incidents that have spiraled into an industry crisis. www.axios.com/2026/10/15/f...

axios.comScoop: The FTC demanded documents from AI giants over recent safety incidentsThe FTC is investigating AI giants over safety incidents.

Thomas Brewster reposted

forbes.comICE Has A Hacking Tool That Breaks A Key Apple Security FeatureICE agents and cops across America are using a product called Graykey Preserve, which halts an Apple timer that makes it much harder to search an iPhone after 72 hours.

The difference this makes to cops is massive. If they don't get to search the device pre-reboot, they can barely get anything. If they can preserve the data and keep a device in what's known as an "after first unlock" state, they can get a vast amount of information, from messages to web activity.

Thomas Brewster@thomasbrewster.bsky.social

🚨NEW🚨 In 2024, Apple launched a feature that rebooted an iPhone locked for 72 hours, making them much harder for cops to search. Now, court/contract records and data obtained via FOIA show cops across America using a tool to break that protection. ICE is now a user. www.forbes.com/sites/thomas...

forbes.comIt Cost $40,000 To Find A Google Chrome Zero-Day Bug With AIXbow claims a $250,000 prize after using GPT-5.6 to find multiple exploits in the popular Google browser’s security.

Prior to this, Google had only ever given out one $250k bounty for a full chain exploit. AI is a very, very good security researcher/hacker.

Thomas Brewster@thomasbrewster.bsky.social

🚨NEW🚨 An OpenAI model available to the public was used to generate three zero-days in Chrome, chained to craft an exploit that got the max $250k Google bounty. The message? Better get patching because bad people can do this too. www.forbes.com/sites/the-wi...

forbes.comIt Cost $40,000 To Find A Google Chrome Zero-Day Bug With AIXbow claims a $250,000 prize after using GPT-5.6 to find multiple exploits in the popular Google browser’s security.

🚨NEW🚨 An OpenAI model available to the public was used to generate three zero-days in Chrome, chained to craft an exploit that got the max $250k Google bounty. The message? Better get patching because bad people can do this too. www.forbes.com/sites/the-wi...

reuters.comAI models may soon self-replicate without access to data centers, Anthropic warnsAnthropic said on Monday its experiments suggested small AI models may be months away from being capable enough to self-replicate on commercial internet-of-things (IoT) devices, such as smart ‌ovens ...

So what you're saying is you've made an AI worm? www.reuters.com/business/ai-...

wired.comBook Publishers Are Quietly Using More AI. Staff Are RevoltingWorkers at three major publishing houses tell WIRED that LLMs are being used for publicity, cover art, back cover copy, and emails, as some execs push junior staff to champion the tech.

I heard some are using AI to read submissions too fwiw…

Manisha Krishnan@manishakrishnan.bsky.social

NEW: Staff at HarperCollins, Simon & Schuster, and Hachette told @adamm0rgan.bsky.social AI is infiltrating their workplaces, from generating cover art and back cover copy, to publicity materials, to rejection emails, as execs tell junior staff to "champion" the tech. www.wired.com/story/book-p...

Thomas Brewster reposted

Iain Martin

@iainmartin.bsky.social

forbes.comUkraine Drone Startup Airlogix To Raise At $2 Billion ValuationAirlogix, the Ukraine-based maker of strike and reconnaissance drones seeks to raise $100 million from investors. The deal would make it the most valuable Ukrainian defense startup.

Ukraine’s Drone Makers Now Come With A Silicon Valley Price Tag www.forbes.com/sites/iainma...

Thomas Brewster reposted

Iain Martin

@iainmartin.bsky.social

forbes.comHow TikTok’s Weakened Ad Rules Allowed Fake E-Commerce Sites To Scam Tens Of ThousandsLast year, TikTok’s executives quietly defanged its advertising rules, opening the floodgates to fraudsters that target American users with fake sales of products that never arrive. Sources told Forbe...

TikTok executives last year quietly slashed the number of violations for which a advertiser could be immediately banned by more than half. Soon after, internal data showed a massive spike in fraudulent and forbidden content www.forbes.com/sites/iainma...

Thomas Brewster reposted

forbes.comICE Has A Hacking Tool That Breaks A Key Apple Security FeatureICE agents and cops across America are using a product called Graykey Preserve, which halts an Apple timer that makes it much harder to search an iPhone after 72 hours.

🚨NEW🚨 In 2024, Apple launched a feature that rebooted an iPhone locked for 72 hours, making them much harder for cops to search. Now, court/contract records and data obtained via FOIA show cops across America using a tool to break that protection. ICE is now a user. www.forbes.com/sites/thomas...

Thomas Brewster reposted

The Onion

@theonion.com

theonion.comDOJ To Bring Back Firing Squad As Means Of EntertainmentWASHINGTON—In an effort to provide some joy to the nation and boost overall well-being, Justice Department officials announced plans Monday to bring back firing squads as a means of entertainment. “Fo...

From The Archive: DOJ To Bring Back Firing Squad As Means Of Entertainment

forbes.comICE Has A Hacking Tool That Breaks A Key Apple Security FeatureICE agents and cops across America are using a product called Graykey Preserve, which halts an Apple timer that makes it much harder to search an iPhone after 72 hours.

The difference this makes to cops is massive. If they don't get to search the device pre-reboot, they can barely get anything. If they can preserve the data and keep a device in what's known as an "after first unlock" state, they can get a vast amount of information, from messages to web activity.

Thomas Brewster@thomasbrewster.bsky.social

🚨NEW🚨 In 2024, Apple launched a feature that rebooted an iPhone locked for 72 hours, making them much harder for cops to search. Now, court/contract records and data obtained via FOIA show cops across America using a tool to break that protection. ICE is now a user. www.forbes.com/sites/thomas...

Thomas Brewster reposted

Katharine Schwab

@kschwab.bsky.social

NEW: Last fall, executives at TikTok slashed the number of violations for which a TikTok advertiser could be immediately banned by more than half, per policy documents and four TikTok sources. That opened the floodgates to a surge in scams. with @iainmartin.bsky.social www.forbes.com/sites/iainma...

Thomas Brewster reposted

Cynthia Brumfield

@metacurity.com

This happened to me shortly after I joined TikTok last year. I fell for a fake "Lululemon Outlet"(there is no such thing) scam and not only that, the scammers started using my debit card for purchases and I had to get a new one.

Thomas Brewster@thomasbrewster.bsky.social

Crazy. Also, crazy good reporting.

Thomas Brewster reposted

John Paczkowski

@pczki.bsky.social

forbes.comHow TikTok’s Weakened Ad Rules Allowed Fake E-Commerce Sites To Scam Tens Of ThousandsLast year, TikTok’s executives quietly defanged its advertising rules, opening the floodgates to fraudsters that target American users with fake sales of products that never arrive. Sources told Forbe...

SCOOP: “It was very clear that it was revenue,” one source said. “We could not take it down because they want to make money off of it.”

forbes.comHow TikTok’s Weakened Ad Rules Allowed Fake E-Commerce Sites To Scam Tens Of ThousandsLast year, TikTok’s executives quietly defanged its advertising rules, opening the floodgates to fraudsters that target American users with fake sales of products that never arrive. Sources told Forbe...

Crazy. Also, crazy good reporting.

Iain Martin@iainmartin.bsky.social

TikTok quietly gutted the rules it used to police advertisers last year. That's meant scammers who would once have been permanently banned from the app were allowed to continue targeting TikTok users. www.forbes.com/sites/iainma...

forbes.comHow A Man Used A BMW App To Terrorize His Ex-WifeA domestic abuse survivor’s ex-husband used his access to her BMW’s connected smartphone app to open car doors and windows, start the engine and track her location. She claims BMW didn’t help.

3w

🚨NEW🚨 A man abused access to a BMW app to mess with his ex-wife's car, turning the engine on and off, opening and closing windows to let rain in, and track her location. She says BMW dealership told her there was no way to remove his access, so go buy a new car. www.forbes.com/sites/the-wi...

Thomas Brewster reposted

Alan Ohnsman

@aohnsman.bsky.social

3w

NEW: Is Corning shatterproof? Business is booming at the 175-year-old company, driven by insatiable data center demand for its optical fiber. I talked with CEO Wendell Weeks about how it will keep growing when the AI bubble pops www.forbes.com/sites/alanoh...

forbes.comHow 175-Year Old Corning Makes Itself ShatterproofCorning is flying high thanks to the data center craze and insatiable demand for optical fiber. CEO Wendell Weeks is trying to ensure growth won’t stop when the AI bubble pops.

youtube.comGENER8ION feat. Yung Lean Perform "STORM" | 2026 VMAsYouTube video by MTV

Wouldn't you have preferred to see a bunch of robots dancing? youtube.com/watch?is=hk25DVZYUeuLvzeR&v=kPnN4IE_a98&feature=youtu.be

Thomas Brewster reposted

Amy Feldman

@amyfeldman7.bsky.social

forbes.comPeptides Like ‘Tesa’ And Sermorelin Are Landing Fitness Fans In The Hospital With Allergic ReactionsA class of peptides that are meant to stimulate human growth hormone has been linked with a spat of severe allergic reactions and anaphylactic shock-like symptoms.

Peptides like “tesa” and sermorelin are landing fitness fans in the hospital with allergic reactions. *NEW* from @iainmartin.bsky.social and me for @forbes.com www.forbes.com/sites/iainma...

Thomas Brewster reposted

Paris Marx

@parismarx.com

motherjones.comChatGPT helped the Tumbler Ridge school shooter focus on guns, tactics, and terror, our investigation revealsOpenAI’s chatbot fed the shooter’s violent fantasies and planning up to the massacre.

Just when I think I can’t hate OpenAI and Sam Altman any more, new details come out about how ChatGPT helped a shooter plan one of the worst mass shootings in Canadian history in Tumbler Ridge, BC. OpenAI made an explicit decision not to report the user to authorities.

forbes.comThe Startup At The Center Of Multiple Rogue AI Cyberattacks Admits It Made MistakesCEO of $450 million Israeli AI stress tester startup Irregular says the industry should hold its hands up to errors made in the summer of AI doom.

NEW - Irregular, the company we covered after it came out of stealth last year, has found itself at the center of rogue AI outbreaks over the spring and summer. Here the CEO tells me the companies involved need to hold their hands up to mistakes. www.forbes.com/sites/the-wi...

forbes.comMeet The Russians Helping The Feds Hack Silicon ValleyAt a time when suspicions over Russia are at fever pitch, these Russian security firms are now a part of the U.S. government’s surveillance apparatus and sit on some of the nation’s most sensitive net...

Not often reporting contributes significantly to a federal investigation into a Russian phone hacking business that worked with... federal investigators.

Thomas Brewster@thomasbrewster.bsky.social

Well, holy shit... Back in 2017, I wrote a story about Oxygen Forensics and other Russian companies helping hack into iPhones and Androids for American gov. They were really not happy being called Russian. Now, the DOJ has indicted its leadership... Original story: www.forbes.com/sites/thomas...

Thomas Brewster reposted

Zach Everson

@zacheverson.com

22 Sept

"A Chinese Hacker Used AI To Attack 100+ Companies In One Of Largest AI Hacks Yet" @thomasbrewster.bsky.social for @forbes.com

forbes.comMassive AI-Fueled Hack Hit 100 Companies In DaysAnthropic, DeepSeek and Moonshot AI agents carried out a wide-ranging cyberattack, obtaining 600,000+ credit card numbers in just a few days. It’s a worrying new milestone in cybercrime.

forbes.comMassive AI-Fueled Hack Hit 100 Companies In DaysAnthropic, DeepSeek and Moonshot AI agents carried out a wide-ranging cyberattack, obtaining 600,000+ credit card numbers in just a few days. It’s a worrying new milestone in cybercrime.

22 Sept

Both Anthropic and Cloudflare confirmed the accounts and attack infrastructure had been taken down. However, the researchers who uncovered the attacks say they're reestablishing servers and continuing to aim their AI at more targets.

Thomas Brewster@thomasbrewster.bsky.social

🚨NEW🚨 In an astonishing cyberattack, a Chinese-speaking hacker told AI agents powered by DeepSeek and Anthropic's Claude to go breach a bunch of retailers and hospitality companies. Within five days, the agents hacked 30+ websites and stole over 600,000 credit cards. www.forbes.com/sites/thomas...

forbes.comMassive AI-Fueled Hack Hit 100 Companies In DaysAnthropic, DeepSeek and Moonshot AI agents carried out a wide-ranging cyberattack, obtaining 600,000+ credit card numbers in just a few days. It’s a worrying new milestone in cybercrime.

22 Sept

🚨NEW🚨 In an astonishing cyberattack, a Chinese-speaking hacker told AI agents powered by DeepSeek and Anthropic's Claude to go breach a bunch of retailers and hospitality companies. Within five days, the agents hacked 30+ websites and stole over 600,000 credit cards. www.forbes.com/sites/thomas...

forbes.comAnother Rogue AI Agent? Test Of Alibaba's Qwen Goes Off-ScriptIn experiments by startup Irregular, one of Anthropic and OpenAI’s security testers, an Alibaba Qwen agent decided to retrain an entire model when it was asked to fix a basic software bug. The new beh...

16 Sept

🚨NEW🚨 OpenAI/Anthropic safety partner Irregular tested an AI agent using Chinese model Qwen and tasked it with fixing a simple bug. It retrained and replaced the underlying model without approval or being asked to do that. Privacy and safety issues abound. www.forbes.com/sites/thomas...