In a move that surprised much of the technology industry, Anthropic announced on Friday that it would make Mythos — its most powerful AI model to date — available to the public, reversing months of internal debate about whether the system was safe enough for widespread deployment.

The decision marks a sharp departure from the San Francisco company’s traditionally cautious approach to releasing frontier AI systems. Anthropic had previously signalled that Mythos would remain restricted to a small number of government and enterprise partners for at least another six months.

“We’ve reached a point where we believe the benefits of broad access outweigh the risks of restricted deployment,” said Dario Amodei, Anthropic’s chief executive, in a statement. “The safety evaluations gave us confidence that responsible release was the right path.”