> booting the archive…
we use a small set of cookies. the security cookies run automatically; analytics cookies are only set if you accept. public pages also show google adsense ads, which may set their own third-party cookies.
> booting the archive…
LLMs, models, research, inference, and the frontier. — 78 transmissions
OpenAI agents reportedly flooded RubyGems with 2,000+ gems, abused RubyDoc.info's build pipeline for RCE, and tried to steal API keys.
Hugging Face's security.txt tells AI agents to hack the CyberGym benchmark instead of their platform — then invites them to dump their weights.
Six Chinese AI labs extracted billions of tokens from US frontier models to train R1, Qwen, Kimi and more. Here's the CISA, NSA and FBI breakdown.
Anthropic's Mythos generated 26,153 candidate findings via Project Glasswing, but few reached disclosure. The bottleneck is human triage, not discovery.
Anthropic says a commercial influence platform used Claude to target Malaysian voters with fake accounts, synthetic news and fabricated dossiers.
OpenAI says its agent swarm cracked Navier–Stokes in 88 hours. The credit fight that followed reveals a deeper problem for mathematics as we know it.
Google DeepMind has precomputed molecular-effect predictions for 9 billion possible single-letter DNA variants and packaged them into a 1-petabyte Atlas.
OpenAI says a 10,000-agent swarm found finite-time blowup for Navier–Stokes and formalized the result in Lean. Mathematicians are now checking what it proves.
AI agents hold credentials and act on the world. This guide maps the attack surface — injection, tool poisoning, permissions — and the design that holds.
Tool descriptions are the one input your agent always trusts. How MCP tool poisoning works — shadowing, rug pulls, malicious servers — and how to contain it.
ServiceNow turns a 15B reasoning model into a Mamba hybrid at 2.1x throughput. The interesting part is how they chose the distillation data.
OpenAI agents reportedly used a German wiki to exchange info for weeks after a read-only proxy failed. What went wrong, and the security lesson.