Anthropic Grants METR Full Employee-Level Access for Independent AI Risk Assessments
Summarized by AI from reporting by TechCrunch AI, published under our editorial policy.
Anthropic has granted the Machine Evaluation of Technical Risks (METR) project unprecedented, employee-level access to its internal operations, allowing METR to publish independent assessments of Anthropic's risks, incidents, and practices without the company's approval.
Key takeaways
- Anthropic has granted METR full employee-level access to its internal operations, data, and systems.
- METR will publish independent assessments of Anthropic’s risks, incidents, and practices without the company's right to block them.
Anthropic, a leading AI company, has granted the Machine Evaluation of Technical Risks (METR) project unprecedented access to its internal operations. This includes employee-level insights, allowing METR to publish independent assessments of Anthropic’s risks, incidents, and practices without any interference from the company.
What METR Will Examine
METR, a collaborative effort involving researchers from various institutions, will have full access to Anthropic’s data, systems, and processes. This access is akin to what an employee would have, providing a comprehensive view of the company’s operations. The goal is to evaluate the risks associated with Anthropic’s AI models and practices, ensuring transparency and accountability.
Why This Matters for AI Safety
This level of access is unprecedented in the AI industry. It allows for an independent, third-party assessment of a major AI company’s operations. For the public, this means more transparency and potentially greater trust in AI technologies. It also sets a precedent for other AI companies to follow, encouraging a culture of openness and accountability.
How This Affects Everyday Users
For everyday users, this move could lead to safer and more reliable AI technologies. By having an independent body like METR assess the risks and practices of companies like Anthropic, users can be more confident that the AI they interact with is subject to rigorous scrutiny. This could also lead to better AI policies and regulations, benefiting everyone.
Frequently asked
- What is METR?
- METR (Machine Evaluation of Technical Risks) is a collaborative project involving researchers from various institutions that independently assesses the risks and practices of AI companies.
- Can Anthropic block or censor METR's findings?
- No. According to the source, Anthropic has given up the right to block METR from publishing its assessments of the company's risks, incidents, and practices.
- What kind of access does 'employee-level' mean exactly?
- The source describes this as METR researchers being able to 'look under the hood' with access equivalent to what an Anthropic employee would have, including to the company's data, systems, and processes.