Anthropic gives outside AI watchdog “employee-like” access
Summary
Anthropic has allowed a nonprofit group called METR to work inside its offices with access to its AI systems. This move aims to improve safety checks on AI models as their power increases, helping prevent risks to society, the economy, and cybersecurity.Key Facts
- Anthropic granted METR, an outside nonprofit evaluator, access similar to that of its own employees.
- METR’s team is embedded in Anthropic’s offices with access to hardware and internal tools.
- The goal is to assess and reduce risks from powerful AI models developed by Anthropic.
- METR does not receive payment from Anthropic and is funded by donations and government bodies.
- Anthropic cannot stop METR from publishing their findings but can request the removal of sensitive data.
- Other companies like OpenAI’s CEO Sam Altman and SpaceX’s Elon Musk support similar external AI audits.
- Critics worry outside evaluators might have conflicts of interest and call for more government rules.
- Anthropic’s safety efforts are part of its plan to prepare for a public stock offering expected this year.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.
