The recently released AI Safety Index evaluated six leading AI companies on their safety measures and risk assessments, with Anthropic earning the highest grade of C. The other companies—Google DeepMind, Meta, OpenAI, xAI, and Zhipu AI—received grades of D+ or lower, with Meta failing entirely. Max Tegmark of the Future of Life Institute, which published the report, emphasizes that the goal is not to shame but to motivate companies to enhance their safety practices, similar to how universities respond to rankings. Tegmark aims to support researchers in AI safety teams by providing them leverage to gain resources and respect within their companies. The Future of Life Institute, focused on preventing negative outcomes from powerful technologies, has previously called for a pause in AI development to establish safety standards, although companies did not comply. The AI Safety Index assessed companies in six areas: risk assessment, current harms, safety frameworks, existential safety strategy, governance and accountability, and transparency. Despite using various sources of public information, most companies did not respond to the report, except Google DeepMind, which stated that their safety measures extend beyond what the index captures. The evaluation involved independent reviewers, including notable figures like Stuart Russell and Yoshua Bengio.

They found that existing safety activities at these companies are ineffective and provide no quantitative safety guarantees. The report particularly criticized companies for their existential safety strategies, as most have not articulated adequate plans for ensuring artificial general intelligence (AGI) aligns with human values. Only Anthropic achieved relatively high scores, notably in addressing current harms, partly due to their responsible scaling policy. However, Tegmark calls for regulatory oversight akin to the FDA to ensure AI products are safe before market release. He warns that companies are trapped in a competitive race that hinders thorough safety testing, advocating for safety standards that prioritize market advantage for companies meeting them first.