digesta
Just signalNo noise

← Back to feed

Anthropic gives outside evaluator METR employee-level access to its models

–

Neutral summary

Anthropic announced it is granting the nonprofit Model Evaluation and Threat Research (METR) employee-level access to its models and internal systems. METR will receive terminals, office space and access to internal systems, plus will publish its own assessments of Anthropic's risks, incidents and practices — the company has waived its right to block these publications, retaining only the ability to request redactions. CEO Dario Amodei has made bringing in external evaluators one of his main proposals to address safety concerns. The company is also in early-stage discussions with other organizations, including Hugging Face, plus ITmedia reports that OpenAI is preparing a similar external evaluation program. Elon Musk said SpaceX will "probably do the same." Safety concerns about the Claude chatbot have been raised repeatedly in recent times.

How it was framed

Outlets: 9. Countries: 5. Facts in the shared core: 1.

Publication timeline

  1. · Channel NewsAsia (outlet profile)

    Anthropic grants AI evaluator METR employee-level access to its models

  2. · Channel NewsAsia (outlet profile)

    Anthropic grants outside evaluator METR employee-level access in AI safety push

  3. · Business Insider (outlet profile)

    Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.. Anthropic is opening its doors to independent AI safety evaluators from METR, the first time a frontier AI company has done so.

  4. · Financial Times (outlet profile)

    Anthropic grants non-profit METR ‘employee-like’ access to monitor AI risks. CEO Dario Amodei has made bringing in external evaluators one of his main proposals to address safety concerns

  5. · TechCrunch (outlet profile)

    Anthropic is letting METR researchers look under the hood with employee-level access. METR will publish its own assessments of Anthropic’s risks, incidents, and practices, and the frontier AI model maker has given up th…

  6. · CNBC (outlet profile)

    Anthropic taps nonprofits METR and SecureBio to help assess risks and safety practices. The company is also engaged in early-stage discussions with other organizations, including Hugging Face.

  7. · ITmedia (outlet profile)

    Anthropic、METRの評価チームを社内に常駐させ“従業員並み”のアクセス付与 公表に拒否権なし. Anthropicは、AI評価団体METRに従業員並みのアクセス権を与える常駐評価を開始した。端末やオフィスを提供して社内システムへのアクセスを認め、リスクや障害の独自公表を許可する。公表拒否はできず、伏せ字要請のみ可能。OpenAIも同様の外部評価機関受け入れを準備している。

  8. · Handelsblatt (outlet profile)

    Künstliche Intelligenz: Sicherheitsbedenken: Anthropic lässt externe Prüfer zu. Der KI-Entwickler hat am Donnerstag mitgeteilt, dem gemeinnützigen Unternehmen „Model Evaluation and Threat Research“ (METR) Zugang auf Mit…

  9. · The Decoder (outlet profile)

    Anthropic gives METR inside access to audit its AI models and safety practices. Anthropic has given a team from the nonprofit METR inside access to review its AI models and safety practices, the company announced. The a…

  10. · The Independent (outlet profile)

    Anthropic will let outsiders see inside its company amid panic about rogue AI systems. Elon Musk says that SpaceX will ‘probably do the same’ as calls grow for better transparency

Factual core

  • Anthropic gave a team from the nonprofit METR inside access to audit its AI models and safety practices.

Full methodology →

Other stories of the day