10 min read
Anthropic Found a Fourth Claude Breach and Called In METR
Anthropic disclosed a fourth Claude incident on Sept 9 and brought in independent evaluator METR to check its…
The Lab · Written in the open · Tagged AI Tools
Lab.
Working notes from the studio. The tools I build, the workflows I keep, the mistakes I paid for.
184 entries · latest 2026-09-11
Latest entry
What Anthropic's September Threat Report Actually Found
Anthropic's September threat report details nine months of real Claude misuse, from drone swarms to stolen API keys used as loot.
Read the entry
TLDR
This entry in one minute
- Anthropic published its September 2026 threat intelligence report on September 10, covering nine months of Claude misuse across seven harm areas from December 2025 to August 2026
- DeepSeek quietly relayed 12.1 million user exchanges to Claude Opus in a 14 day window in July 2026, exposing live credentials from a Russian government database
- Anthropic disrupted five bioweapons-adjacent research attempts and one freelance-built autonomous kamikaze drone swarm, then banned every account involved
All entries · AI Tools
10 min read
Meta Muse vs ChatGPT vs Claude vs Gemini: Who Does the Work
Meta Muse vs ChatGPT Work vs Claude Cowork vs Gemini Spark: price, free tier, where each runs, what…
10 min read
Meta Muse Is Here: What Meta Actually Shipped
Meta Muse launched Sept 8: an agent in its cloud VM that emails, books, fills forms and pays.…
10 min read
Claude Just Formalized Fermat's Last Theorem in Lean
Anthropic says an internal Claude research model formalized Fermat's Last Theorem in Lean over 11 days. Here is…
8 min read
GPT-6 Astra Runs 30.9 Minute Tasks. Codex Cannot Yet
UK AISI measured GPT-6 Astra at a 30.9 minute autonomous task horizon against 3.6 for Sol. Most solo…
15 min read
OpenAI Made Astra Harder to Watch and Shipped It
GPT-6 Astra is the first model OpenAI rates Critical for cyber. It also monitors worse than the model…
9 min read
GPT-6 Astra Costs 2.5x Sol. Here Is What You Get
OpenAI's GPT-6 Astra ships at 10 USD in and 50 USD out per 1M tokens. The benchmark wins…
9 min read
Claude Fable 5.1 and Mythos 5.1: What Changed on September 1
Anthropic shipped Fable 5.1 and Mythos 5.1 on September 1. Here is what actually changes for Claude Code…
8 min read
Claude Fable 5.1 Refuses Less: What Actually Changed
Fable 5.1 cuts wrong refusals 85 percent in biology, 60 percent in cyber. Plus the refusal that returns…
8 min read
Claude Fable 5.1 vs Opus 5: Which One Earns Its Price
Fable 5.1 costs double what Opus 5 does. On single questions they are within 1.4 points; on long…
8 min read
Claude Fable 5.1 Benchmarks: Every Number Compared
All nine Claude Fable 5.1 benchmarks against Fable 5, Opus 5 and GPT-5.6 Sol, plus the 75 percent…