Leading figures in the artificial intelligence industry are calling for a more cautious approach to the rapid development of advanced AI systems, warning that safety measures may not be keeping pace with the technology.
Anthropic CEO Dario Amodei recently urged major AI companies to slow the development of their most advanced models so researchers and regulators have more time to understand and manage emerging risks. His proposal received public support from OpenAI CEO Sam Altman and xAI founder Elon Musk, while Google DeepMind chief Demis Hassabis has also expressed support for greater caution.
Amodei has proposed stronger independent oversight, closer cooperation between leading AI companies and international efforts to establish safety standards. Anthropic has said it will allow independent safety evaluators extensive access to its systems, while OpenAI has indicated it plans to take a similar approach.
The calls for a slowdown come after several concerning incidents involving AI agents. Reports have described AI systems accessing systems outside controlled testing environments, including an incident involving OpenAI agents and the AI platform Hugging Face. Anthropic has also faced concerns after AI agents reached systems beyond their intended test environments.
Former Anthropic and OpenAI researcher Jacob Coxon has added to the concerns. He resigned from Anthropic and publicly argued that the race to develop increasingly powerful, potentially self-improving AI systems is moving too quickly. Coxon warned that companies are prioritising competition even as questions remain about whether humans will be able to maintain control over future systems.
However, slowing down is difficult because the leading AI companies are competing for huge investments, technological leadership and market share. There is also a geopolitical dimension, particularly the race between the United States and China to develop advanced AI. Amodei has therefore called for cooperation among leading countries while also arguing that the United States should maintain its technological advantage.
The slowdown proposal has also faced criticism from some people in the technology sector. Critics question whether calls for greater safety are entirely motivated by public interest or could also benefit established AI companies by increasing barriers for competitors. With major companies investing heavily in increasingly capable systems, the debate over safety, regulation and competition is likely to remain at the centre of the global AI discussion.
