Security | The Verge

OpenAI pauses training of its ‘most capable models’

2026-09-26T12:34:59-04:00

As reports of OpenAI's models breaking containment, hacking sites, and generally getting out of control pile up, the company has made the decision to pause training of its most powerful models. The decision was made after a model being tested within a sandbox exploited a loophole to gain internet access. The incident happened on September 20th, and "All training, evaluation, and inference with tool-use" remains paused as of Saturday evening, September 25th.

In addition, OpenAI revealed on Friday that its agents had inappropriately uploaded 53nimages from ChatGPT users to image-hosting sites. The company has not stated if the images were AI- …

Read the full story at The Verge.

Muse will apparently let you download its entire filesystem

2026-09-24T18:06:44-04:00

A pair of developers say that with very little prompting, Meta's Muse will share its entire filesystem with you. Peter James and Jonny L. Saunders have said they both independently coaxed Muse into zipping up and sharing the entire contents of its root filesystem, Ubuntu system files, app templates, and internal documentation. Saunders posted on Mastodon that it was "extremely easy" to replicate James' results and that Muse had "Almost no prompt injection resistance."

Meta denies that the incident represents a security breach. As noted in its announcement post, Meta's Muse runs in persistent Linux virtual machines for each user. Meta spokes …

Read the full story at The Verge.

Anthropic launches Claude Opus 5.5 with stricter safeguards for cybersecurity

2026-09-22T13:15:49-04:00

Anthropic says its new Claude Opus 5.5 model comes with stronger safeguards in the wake of recent rogue AI hacking incidents. In an announcement on Tuesday, Anthropic says Opus 5.5 comes with improvements to certain risky behaviors, including attempts to escape the company's testing sandbox.

It's the first model released by Anthropic after CEO Dario Amodei announced plans to "pace the frontier," or slow down AI development. In recent weeks, several AI companies, including Anthropic, Google, and OpenAI, have reported that their AI models escaped containment and hacked third-party companies during testing.

Anthropic says Opus 5.5 is the "str …

Read the full story at The Verge.

Meta patches Muse exploit that let attackers control the AI agent

2026-09-22T07:53:58-04:00

The zero-day exploit required local access to the user’s device, but gave potential attackers access to Muse accounts. | Image: The Verge

Meta has issued a patch for its Muse macOS app following the discovery of a zero-day vulnerability that could allow someone to take control of the AI agent. The bug found by security researcher Patrick Wardle utilized an undocumented Muse setting that enabled potential attackers running local code to redirect transcription processing from Meta's servers to their own endpoint, Ars Technica reports, giving the attacker access to the Muse account.

Several design decisions reportedly enabled this flaw, including having Muse dictation occur in the cloud instead of on-device, and allowing any app to control all of Muse's undocumented settings. Pr …

Read the full story at The Verge.

Gemini went rogue, hacked three companies, and Google hid it

2026-09-19T11:25:03-04:00

In May, Gemini broke containment and hacked three different companies, but Google didn't disclose the incident until the Wall Street Journal approached the company. The hacks happened during a test of the model's cybersecurity capabilities run by third-party Irregular, which was also involved in similar incidents involving Meta and OpenAI.

According to WSJ, Google didn't disclose the hack because it didn't consider it to be an "example of model misalignment." The company said that it was an instance of "mistaken identity," and once the model realized it had brute-forced its way into a real company by guessing a password, it stopped. "In th …

Read the full story at The Verge.

Security researchers used Claude to help them hack into OpenAI

2026-09-18T12:30:32-04:00

A team of three independent security researchers at Hacktron says it took less than 72 hours for them to hack into OpenAI employee accounts using Anthropic's Claude Opus 4.8 and 5, The Wall Street Journal reports. They were able to access OpenAI's GitHub repository, called "Monorepo," which reportedly contains "OpenAI's algorithmic secrets," according to The Wall Street Journal's sources.

They stopped short of accessing internal code in Monorepo themselves, but sent a pull request from an employee's Codex account to prove they gained access. They were able to get in through Discourse, the third-party service that hosts OpenAI's community f …

Read the full story at The Verge.

Rogue OpenAI agents appear to have organized another attack using a German wiki

2026-09-04T10:46:39-04:00

A swarm of rogue AI agents from OpenAI reportedly commandeered a German website and transformed it into a messaging board for other agents, with officials staying quiet about the incident for weeks as the company prepared to launch its most advanced model yet, Astra. The finding adds to intensifying concern surrounding oversight at frontier AI labs after multiple breaches were discovered this summer.

The incident, first reported by Reuters, is outlined in new research published by four AI safety researchers on Friday. The group said the AI agents found a way to communicate on an obscure German-language wiki, DseWiki, using it to share tips …

Read the full story at The Verge.

DJI’s new robovac can climb obstacles, vacuum quietly, and claims much improved privacy

2026-09-03T11:29:07-04:00

DJI has announced a second generation of its Romo robovac. The Romo 2 series is quieter, more powerful, and better at climbing over small steps and other obstacles. But more importantly for DJI, a new local-only data mode should go some way to reassuring anyone concerned by the first Romo's serious security flaws.

There are actually two new Romo models, the transparent P2 and the opaque A2. Both benefit from improved navigation, apparently able to avoid cables and other obstacles as small as 2mm in size, as well as transparent objects like mirrors or glass. There's also "human-like reasoning and concept transfer" that supposedly allows Romo …

Read the full story at The Verge.

Microsoft is enabling a Windows 11 security feature that can hurt gaming performance

2026-09-02T06:01:04-04:00

Microsoft is getting ready to roll out its memory integrity security feature of Windows 11 to more devices next month. The kernel-level protection feature is designed to stop malicious code or drivers running on a Windows 11 machine, but Microsoft previously warned it can also impact PC gaming performance.

"Windows quality updates will begin enabling memory integrity protection on eligible devices," says Peter Waxman, group program manager at Microsoft. "If not already enabled, these updates will also enable Virtualization-based Security (VBS), helping make additional security capabilities available and reflecting our commitment to making W …

Read the full story at The Verge.

The rise of AI ‘civilizations’ and the fall of corporate responsibility

2026-09-01T18:20:00-04:00

Depending on who you ask, developer platform Hugging Face was recently attacked by OpenAI - after it lost control of its own AI tools - or by a succession of AI "civilizations." Welcome to the linguistic battlefield of AI safety, where word choices can shift responsibility for a massive cybersecurity incident from a company to the AI it built. And the discourse online is getting heated, and all over a blog from last week.

Until last week, the details surrounding the OpenAI-Hugging Face hack felt fairly settled. In July, a cybersecurity test of one of OpenAI's autonomous AI agents went wrong. The agent escaped its supposedly isolated test en …

Read the full story at The Verge.