Leading artificial intelligence companies are unprepared for the risks their own models pose, two major reports argued.
The Future of Life Institute and Safer AI, two respected nonprofits, rated top AI firms on how well they identified and prepared for potential catastrophic risks, such as bio- or cyber-terrorism.
Both groups said that even the most well-prepared companies had worrying gaps on existential risk in their plans, and FLI said that “none of the companies has anything like a coherent, actionable plan” for controlling increasingly powerful systems. Anthropic scored highest on both reports, but that is damning with faint praise: It received a C+ grade from FLI and 35/100 from Safer AI.