AI Companies Lack Plans for Managing AGI Risks, Report Finds

AI Companies Lack Plans for Managing AGI Risks, Report Finds
……………….

Artificial intelligence companies are unprepared to manage the risks associated with developing human-level AI, according to a report by the Future of Life Institute (FLI). The study found that none of the seven leading AI firms evaluated scored higher than a D in “existential safety planning.”

The report assessed Google DeepMind, OpenAI, Anthropic, Meta, xAI, Zhipu AI, and DeepSeek across six areas, including current harms and long-term safety. Anthropic received the highest overall grade with a C+, followed by OpenAI (C) and Google DeepMind (C−).

Reviewers noted that despite pursuing artificial general intelligence (AGI)—AI systems capable of performing any intellectual task a human can—none of the companies presented a clear plan for ensuring control or safety. The FLI warned that without effective risk management, AGI could pose existential threats.

A separate report by SaferAI, another nonprofit focused on AI safety, described the industry’s current practices as “weak to very weak” and unacceptable. FLI co-founder Max Tegmark compared the situation to building a nuclear plant without safety protocols. Google DeepMind responded that the evaluation did not fully reflect its broader AI safety measures.