AI safety push exposes internal tensions at OpenAI and Anthropic
Staff at OpenAI and Anthropic are reportedly debating how calls to slow AI development could be translated into practical measures.
Employees at OpenAI and Anthropic are grappling with how to turn calls for a slowdown in artificial intelligence development into practical measures, creating internal tensions at the Silicon Valley companies, the Financial Times reports.
The supplied report does not detail the participants, proposals or specific disagreements behind the debate. Nor does it establish the scale or consequences of the reported tensions.
The discussions come after the companies considered voluntary standards for frontier AI safety with other industry participants, including Google DeepMind. The proposed framework has not been presented as a confirmed agreement.
Anthropic chief executive Dario Amodei has said the company was discussing stronger safety standards and checks across the industry. OpenAI executive Chris Lehane has said OpenAI, Anthropic and Google DeepMind were working on voluntary frontier-AI standards.
It remains unclear whether those measures will become formal or enforceable, leaving the companies to contend with how broad safety ambitions might be applied in practice.
