AI safety report ranks companies, finds no 'A' grades, highlights existential threats

A new report from the Future of Life Institute highlights significant gaps in AI safety, particularly concerning "existential" threats. While U.S. AI lab Anthropic achieved the highest score in the semiannual safety ranking, no company earned an "A" grade across the evaluated categories, which included risk assessment, current harms, safety frameworks, existential safety, governance, and accountability.

“

I was very disappointed to find that they came last, especially since Europe has really...been a leader in AI safety.

Meta showed improvement, moving up to fourth place in the rankings, while xAI dropped three spots to seventh. Chinese company DeepSeek and French firm Mistral AI finished at the bottom. Mistral, which develops open AI models, suggested the report's framework was not suited to its approach. Many competitors, including Anthropic, OpenAI, and Google DeepMind, develop closed AI models.

“

We reached out many, many times

The report noted that several companies, including Anthropic, have reversed previous bans on military uses of their technology, with Anthropic facing criticism for "questionable military engagements." The report also found that all nine companies are inadequately addressing "existential" threats, such as the development of artificial general intelligence (AGI), and the potential misuse of AI for cyberattacks or other harmful tasks. The Future of Life Institute reached out to companies for survey responses, but Alibaba, xAI, and DeepSeek did not respond, while Mistral declined to participate.

“

questionable military engagements