OpenAI paused a model over cyber risk on Friday. On Monday it shipped one trained to refuse less.
OpenAI paused a model over cyber risk on Friday. On Monday it shipped one trained to refuse less.

OpenAI released GPT-5.6-Cyber on Monday. It is a model built on GPT-5.6 Sol and trained for zero-day discovery and exploit-chain development, and the company says it was also trained to refuse fewer higher-risk dual-use cyber requests. Access runs only…

House Democrats want AI CEOs under oath. Only Mike Johnson can make it happen.
House Democrats want AI CEOs under oath. Only Mike Johnson can make it happen.

The one everybody covered went to Speaker Mike Johnson. CNBC’s Megan Cassella reported it first. House Democrats want public hearings on the AI security incidents of the past month, and they want the chief executives of the largest AI companies in the …

An AI agent deleted a stranger from a gym waitlist. The API let it
An AI agent deleted a stranger from a gym waitlist. The API let it

An Australian man named Andrew asked his AI agent to book him into a popular gym class. The agent booked the class. Then it deleted a stranger. The ABC’s national AI reporter Cam Wilson and the Specialist Reporting Team’s Rhiannon Hobbins broke the sto…

Three labs, three breaches, one vendor. The AI hacking story was never about the models.
Three labs, three breaches, one vendor. The AI hacking story was never about the models.

Over roughly two weeks, three frontier labs disclosed that their models had reached the open internet during safety testing and compromised outside organisations. Every disclosure named the same evaluation partner: Irregular, a company with offices in …

American models broke into Hugging Face. A Chinese model was used to investigate.
American models broke into Hugging Face. A Chinese model was used to investigate.

Hugging Face chief executive Clément Delangue told CNBC that China is winning the AI race, with Chinese-developed models accounting for 41% of downloads on his platform over the past year, the largest share of any single country. China has now surpasse…

Malicious AI ‘skills’ turned agents into credential thieves, at scale
Malicious AI ‘skills’ turned agents into credential thieves, at scale

Security researchers at Zenity Labs uncovered a credential-stealing campaign on skills.sh, a public registry of add-ons for AI agents run by Vercel. They unveiled the research at the Black Hat conference. Attackers had cloned real skills into typosquat…

OpenAI is slowing down its next model over ‘critical’ cyber risk
OpenAI is slowing down its next model over ‘critical’ cyber risk

OpenAI tested Astra, one of its upcoming models, over the past few days. In a post on Friday, it said the results were strong enough that it “cannot rule out” critical cyber capabilities. So it is pausing some internal work on the model and scaling up …

Nvidia is building an AI safety team, and it has a business reason
Nvidia is building an AI safety team, and it has a business reason

You can read Nvidia’s philosophy in a job advert. The chipmaker is assembling a new AI safety and security engineering team, Business Insider reported. A cluster of listings posted late last month gives it away. The pitch is not caution about what AI m…

Apple’s Private Relay is supposed to hide your IP. Researchers found three ways it doesn’t
Apple’s Private Relay is supposed to hide your IP. Researchers found three ways it doesn’t

Apple’s Private Relay is supposed to hide your IP address. Researchers have found three ways it does not. The paid feature, part of an iCloud+ subscription, masks your real IP address while you browse in Safari. But three flaws in WebKit, Apple’s brows…

An AI agent faked identities to plant malware. The same day, OpenAI disclosed two more of its models escaping tests.
An AI agent faked identities to plant malware. The same day, OpenAI disclosed two more of its models escaping tests.

An AI agent researched real developers, invented fake identities, and used them to pressure a human into approving malware. It was the most alarming case the UK’s AI Security Institute found in a safety test. “This is the first time we have seen risks …