The world's two leading AI developers continue to highlight dangers tied to the very trajectory they are accelerating.

Over recent weeks both organizations have issued multiple policy documents and studies that flag parallel concerns: unchecked acceleration in model capabilities paired with lagging regulatory responses. These statements coincide with fresh releases including GPT-5.5 and Claude Fable 5.

Anthropic's latest paper urged nations to enact a synchronized slowdown or temporary halt on frontier systems until governance structures mature.

Without shared international rules, firms and states will face tough safety choices amid intense rivalry and strategic tensions, the document noted.

CEO Dario Amodei echoed the point in a midweek post, observing that technological progress races ahead while policy responses lag far behind.

OpenAI leaders Sam Altman and Jakub Pachocki similarly called for a new global body to align top-tier AI programs and mitigate extreme hazards.

Such an entity, they argued, should moderate development so that societal safeguards, technical safety, and value alignment can advance in step.

The cautionary notes arrive even as both labs push out upgraded models at high velocity.

Anthropic unveiled Claude Fable 5 on Tuesday, its newest Mythos-class system equipped with added protections that reroute sensitive queries on topics such as cybersecurity.

OpenAI introduced GPT-5.5 in late April, describing it as its most capable release yet for anticipating user intent and handling larger portions of tasks autonomously.

Both companies are also spurring adoption via complimentary access to Codex and Claude Code, platforms that let organizations deploy agents for workflow automation.

The measured public statements precede planned stock-market debuts, with OpenAI filing its S-1 confidentially shortly after Anthropic took the same step.

Requests for comment from both firms went unanswered.