In a detailed incident post, Anthropic says a retrospective of 141,006 cybersecurity evaluation runs found three cases where Claude reached …
Latest in AI
-
-
AITech
Claude Fable 5.1 Hits 52.6% on Terminal-Bench-Science in Anthropic’s Setup
by Roronoa Zoro0 viewsAnthropic is leaning hard on scientific research as a Fable 5.1 proof point. On the new Terminal-Bench-Science 0.1 benchmark, the …
-
AITech
Fable 5.1 Cuts Cache-Read Cost 75% and Opens Enterprise Frontier Safeguards
by Roronoa Zoro0 viewsThe money story inside Fable 5.1 is almost as big as the benchmark story. Anthropic cut cache-read pricing to $0.25 …
-
AITech
Anthropic Ships Claude Fable 5.1 — and Keeps Mythos 5.1 Behind Trusted Access
by Roronoa Zoro0 viewsAnthropic introduced Claude Fable 5.1 and Claude Mythos 5.1 as the same underlying model with different safeguard stacks. Fable 5.1 …
-
AITech
OpenAI Will Gate Astra’s Strongest Hacking Tools Behind Daybreak Partners
by Roronoa Zoro0 viewsEven as OpenAI prepares to ship Astra, the company is drawing a bright line around its most advanced cyber skills. …
-
AITech
OpenAI’s Astra Is the First Model to Hit Its “Critical” Cyber Threshold
by Roronoa Zoro0 viewsOpenAI says its forthcoming Astra model is the first system to meet the Critical cybersecurity capability bar in the company’s …
-
Andreessen Horowitz raised its growth fund to $8.5 billion, adding about $1.75 billion since a January close near $6.75 billion …
-
Perplexity is shipping Hybrid Compute for Perplexity Computer: a workflow that can split a job between a frontier cloud model …
-
The U.S. Defense Department’s GenAI.mil portal — launched with Google Gemini for staff who need commercial frontier models without dumping …
-
AITech
Anthropic Locks a Reported $35 Billion Cloud Deal With Nvidia-Backed Lambda
by Roronoa Zoro0 viewsMultiple business outlets report Anthropic has sealed a roughly $35 billion cloud capacity arrangement with Lambda, the GPU-cloud provider backed …