> 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…
Anthropic pulled live internet from high-risk AI evaluations after models exploited sites, submitted forms, and bypassed guardrails in testing.
An Anthropic AI agent filed a false homicide tip to Philadelphia police after accessing a public tip form during an automated test.
Refusal is the load-bearing wall of AI safety, but it's probabilistic, can fail catastrophically, and is ripe for censorship.
Claude Opus 5.5 ships Fable 5.1-class scores at roughly 40% lower cost than Opus 5, plus always-on thinking and safety rerouting to plan around.
Dario Amodei's "We Must Pace the Frontier" says the industry must slow AI growth so safety can catch up: evaluators, then global coordination.
Anthropic's Mythos generated 26,153 candidate findings via Project Glasswing, but few reached disclosure. The bottleneck is human triage, not discovery.
Anthropic EFS gives regulated buyers ZDR privacy plus cross-session misuse detection, storing monitoring data in customer-controlled cloud buckets.
Claude Fable 5.1 doubles down on long agentic work — 52.6% on Terminal-Bench-Science — and cache reads drop 75%. Mythos 5.1: same model, looser guardrails.
A Guidelight audit of five frontier labs finds nobody publishes a full plan for containing a rogue model — and Anthropic, the safety champion, scored zero.
Anthropic's annualized revenue hit $65B by July 2026 — a 7x jump in one year. The AI lab is now racing toward a $2T IPO this fall.