What scale did Anthropic use to rate AI's involvement?

It used the six-level "AL0" to "AL5" scale created by the US research organization Epoch AI. AL0 means no AI involvement, AL3 means working with AI, AL4 means AI-led, and AL5 means AI is autonomous.

How closely does Anthropic monitor its AI agents?

On its most-used agent platform, about 30,000 agents work on research and development at any time as of August. Every action is checked before execution; in August, over 1 billion decisions were analyzed, and the monitoring system stopped 0.002% of them.

Large Language ModelsAI Safety & AlignmentAI Business & IndustryITmedia AI＋Sep 18, 2026Sep 18, 2026, 19:01 JST

Anthropic: Claude leads 26% of its AI R&D work

Send on LINE

3 Key Points

What happened
Anthropic Institute said that as of August, 26% of its AI research and development work was almost entirely done by Claude with humans only supervising. In February it was under 1%.
Why it matters
Adding work done jointly with detailed human direction, the share of R&D touched by AI tops 90%, which is likely why Anthropic published these figures as a transparency gauge.
What to watch
No work reached the fully autonomous "AL5" stage as of August, so the shift hinges on whether that top level stays empty.

WHO IT HITSOther AI labs and their research leads now face a visible benchmark for how much of their own R&D an AI model can run, and safety and oversight teams get a concrete monitoring rate (0.002% of over 1 billion checks stopped) to compare against.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Get these free every morning

FAQ

What scale did Anthropic use to rate AI's involvement?: It used the six-level "AL0" to "AL5" scale created by the US research organization Epoch AI. AL0 means no AI involvement, AL3 means working with AI, AL4 means AI-led, and AL5 means AI is autonomous.
How closely does Anthropic monitor its AI agents?: On its most-used agent platform, about 30,000 agents work on research and development at any time as of August. Every action is checked before execution; in August, over 1 billion decisions were analyzed, and the monitoring system stopped 0.002% of them.

ITmedia AI＋Read Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.