United Kingdom, April 20 -- The US government and Anthropic have held "productive and constructive" discussions.

The meeting took place shortly after Anthropic released its Claude Mythos preview, which can supposedly outperform humans at some hacking and cyber-security tasks.

The White House said in a statement: "We discussed opportunities for collaboration, as well as shared approaches and protocols to address the challenges associated with scaling this technology."

The US government also revealed that the two parties - which have previously collaborated on high-level government and military work - "explored the balance between advancing innovation and ensuring safety".

By contrast, US President Donald Trump previously hit out at Ant...