Policy·
Trump dropped the mandatory AI model review after Silicon Valley pushed back
Trump's June 2 AI executive order asks for a voluntary 30-day model review, down from a mandatory 90-day one. Here's what got cut and who pushed.
#ai-security
RSS
AI·
Anthropic's Glasswing logged 10,000 vulnerabilities in a month. Most are still waiting on a patch.
Anthropic says Project Glasswing's first month produced over 10,000 critical-and-high-severity vulns. Verification and patching is the limiting step.
Policy·
Sadiq Khan blocked a £50M Met Police deal with Palantir. Scotland Yard had only talked to one supplier.
London's mayor cited a 'clear and serious breach' of procurement rules and stopped the Metropolitan Police from awarding Palantir a £50M AI intelligence contract on May 21.
Security·
A bad command-line parser turned every claude-cli:// link into a remote shell
Joernchen of 0day.click found a deeplink RCE in Claude Code. Anthropic shipped the fix in 2.1.118 the same week.
Security·
F5 patched an 18-year-old NGINX bug. Attackers can RCE a third of the web with one crafted request.
F5 disclosed CVE-2026-42945 on May 13 after depthfirst's analyzer found a heap overflow in a 2008 commit. NGINX 1.31.0 ships the patch, every Plus tier needs an upgrade.
Security·
A crafted Ollama model file leaks the whole server's memory. 300,000 instances are exposed.
Cyera disclosed CVE-2026-7482 on May 1, a CVSS 9.1 unauthenticated heap read in Ollama. Three API calls dump prompts, env vars, and API keys from any open instance.
AI·
Chinese proxy networks sell Claude API access at 90% off. They harvest every prompt that passes through.
A ChinaTalk investigation reveals how 'transfer stations' resell Anthropic API access using stolen credentials, model substitution, and prompt harvesting.
Policy·
A judge killed DOGE's grant purge. The 'review process' was asking ChatGPT 'Is this DEI?'
A federal judge restored $100M+ in grants after two DOGE staffers used ChatGPT to flag 97% of NEH grants as DEI, including an HVAC repair and Holocaust research.
Security·
380,000 vibe-coded apps are sitting on the open web. 5,000 of them are leaking real data.
RedAccess found that AI coding tools like Lovable, Base44, and Replit default to public hosting, leaving medical records, bank internals, and corporate secrets indexed by Google.
Security·
Five Eyes intel agencies publish first joint agentic AI security guide. Their advice: slow down.
CISA, NSA, GCHQ, ASD, CSE and NCSC-NZ jointly tell organizations agentic AI isn't ready for fast rollout. The 23-page guide names five risk categories.
Security·
A malicious GGUF file owns your SGLang server: CVE-2026-5760 is an unpatched 9.8
SGLang's reranker renders chat templates without a sandbox. Load a hostile GGUF, hit /v1/rerank, and the attacker has Python on your inference box. No patch yet.
AI·
A Discord group guessed Anthropic's URL pattern and walked into Claude Mythos
Bloomberg reports a small group accessed Anthropic's locked-down Mythos model the same day it launched, using credentials from a third-party contractor and educated URL guessing.
Security·
Malicious npm and PyPI packages turn dev servers into Chinese LLM proxies
Aikido found a stage-2 Go binary inside two health-check-themed packages that runs an OpenAI-compatible router routing Claude, GPT, and Gemini traffic through Chinese aggregators.
Open Source·
Linux 7.1 is yanking ham radio, ISDN, and ATM. The reason: AI bug-report spam.
Jakub Kicinski's networking pull request removes 138,161 lines of decades-old code. Kernel maintainers say LLM-generated bug reports made the old subsystems un-maintainable.
Open Source·
Mozilla fixed 271 Firefox bugs that Claude Mythos found. Its own tests caught 22.
Firefox 150 shipped Monday with 271 security fixes from Anthropic's Project Glasswing. Mozilla CTO Bobby Holley says Mythos matches elite human researchers.
AI·
OpenAI launches GPT-5.4-Cyber for defensive security, opens access to thousands
OpenAI's new cybersecurity-tuned model can reverse-engineer binaries and analyze malware. It's restricted to verified defenders through the Trusted Access program.