AI leaders tout safety concerns as a strategy to secure market dominance

Anthropic and OpenAI CEOs have warned that their cutting-edge AI models could threaten humanity and called for regulation and independent testing, even addressing the United Nations. Experts say the companies are leveraging these warnings to earn public trust and influence safety rules that will benefit them as they prepare for IPOs and a politically charged midterm season. Both firms have highlighted internal safety steps, such as OpenAI’s pause on training its most advanced models, and suggest external auditors should assess progress if government oversight remains limited.

Critics note the focus on speculative existential risks diverts attention from immediate safety issues like AI use in military applications and data-center emissions. Meanwhile, the Trump administration shows little interest in new AI regulations, while the Biden-created U.S. Center for AI Standards and Innovation continues to evaluate models on a voluntary basis. Analysts like Harrison Rolfes view the safety narrative as a way for the leading labs to create a protective barrier that deters smaller rivals and attracts investors.