Guidelight's First AI Company Safety Scorecard Gives Anthropic and OpenAI Both a C+
Guidelight released its first AI company safety scorecard, based on hundreds of hours of reading public documentation, evaluating whether frontier AI companies — Anthropic, OpenAI, Google, xAI and Meta — can effectively control their models. The industry-wide finding is that control practices fall broadly short, with Anthropic and OpenAI tied at just C+. The report aims to serve as an authoritative resource for journalists and policymakers.
Confirmed
- The five companies assessed were Anthropic, OpenAI, Google, xAI and Meta.
- Six foundational practices were evaluated: logging, monitoring effectiveness, gated operations, kill switches, third-party review, and containment plans.
- Anthropic and OpenAI both scored C+; frontier AI companies' control capabilities are generally inadequate.
- Based solely on public materials, the report translates complex safety standards into clear, readable metrics.
Why it matters
- The authors argue companies should practice actual prevention rather than after-the-fact detection, predicting a string of safety incidents until prevention is taken seriously.
2026-08-19 ~ 2026-08-19 · 9 related posts
Primary sources
- Guidelight releases scorecard on AI companies' safety practices — sjgadler · 2026-08-19
- [source] Report: Frontier AI Fails Basic Control Practices; Anthropic and OpenAI Lead with C+ — sjgadler · 2026-08-19
- [source] Guidelight launches first scorecard on AI companies' safety control practices — sjgadler · 2026-08-19
- Opinion: AI firms must shift from detection to actual prevention — sjgadler · 2026-08-19
5 near-duplicate retellings: sjgadler · sjgadler · sjgadler · Miles_Brundage · AccBalanced