Anthropic says unreleased model built working exploits for Chrome, Safari, Windows, Linux
Anthropic reported that an unreleased internal AI model identified security flaws in major operating systems and browsers, chaining them into working proof-of-concept exploits. The company stated the model required only initial prompting and little human involvement to create these attacks, a capability it claims was previously limited to specialist teams. All identified flaws were reported to…
Key points
- Anthropic's unreleased model chained flaws into working exploits for Chrome, Safari, Windows, and Linux.
- Google and Apple credited Anthropic and Claude for specific security fixes in their release notes.
- The Linux exploit achieved root access over the network, linked to a 9.8-rated kernel CVE.
The report details five distinct exploit chains. The Chrome chain used four flaws to escape the sandbox and run code, with Google crediting Anthropic for 62 fixed bugs. The Safari chain disabled the same-origin policy to steal data without executing code. The Windows chain caused a denial-of-service via a kernel bug, while the Linux chain achieved root access over the network. Apple and Google publicly credited Anthropic researchers and its Claude model for specific fixes.
Skepticism remains regarding the degree of human assistance, as Anthropic did not name the specific model or detail human steps. The Linux exploit is linked to a high-severity CVE, though public records describe the fix differently than Anthropic's post. Anthropic frames this as a defensive measure, urging security teams to use its Cyber Verification Program to access advanced models for vulnerability detection.
Model pages: Claude Opus 5.5 → · Sonnet 5.5 →
Anthropic's Unreleased AI Model Chains Flaws Into Working Chrome, Safari, Windows and Linux Exploits
thehackernews.com · 28 October 2026
Loading the full article…
This text was published by thehackernews.com and written by The Hacker News. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Artificial Analysis releases Intelligence Index v5 with private datasets · 1 src
- DeepSeek’s R1 model with guideline knowledge base improves junior doctors’ arthroplasty decisions · 1 src
- Google DeepMind releases Genie 4 world model with native audio and Gemini 3.8 Flash guidance · 12 src
- Researchers propose synthetic document finetuning to detect alignment-faking LLMs · 3 src
- Researchers introduce LOTUS method to detect LLM hallucinations via low-rank decomposition · 1 src
Comments
via GitHub Discussions