Anthropic update on independent evaluators
What happened
Anthropic announced that it is giving a team from the nonprofit METR employee-like access to its internal tools, hardware, and models so they can conduct deep risk assessments and publish their findings. The company also works with SecureBio and Accenture as evaluators and is in early talks with other groups such as Hugging Face. METR’s analysts will receive access comparable to Anthropic’s own risk teams, except for clearly defined confidential categories, and they can publish their analysis without Anthropic veto, though redactions may be required. This move follows Anthropic’s earlier pledge for stronger third‑party safety reviews and its signing of the White House Accord on Super Intelligence.
Why it matters
It shows AI firms are moving toward more transparent safety oversight, which could help build public trust in powerful AI systems. For ordinary readers, it means there may be more independent scrutiny of AI risks rather than solely internal company reports.
Reality check
An automated, sceptical read of the sourceIncremental. The announcement expands existing third‑party evaluation efforts but does not introduce fundamentally new technology or safety mechanisms.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from Anthropic
Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-FreeAnthropic · 14 Nov 2026Claude designed FPGA scaler board for 6th-gen consolesAnthropic · 13 Nov 2026I asked Claude Fable 5.2 to turn my site into a 90s Silicon Graphics workstationAnthropic · 13 Nov 2026Anthropic is endorsing SB 53Anthropic · 13 Nov 2026Understanding and addressing AI harmsAnthropic · 13 Nov 2026This page was collected from anthropic.com and summarised automatically by AIPulsee's pipeline. See how it runs.