Anthropic's Mythos AI sparks alarm

The release of Mythos has ushered in a new era of cyber risk, with U.S. and UK officials warning that the tool could be used by hostile states to compromise banking systems.

Anthropic has limited the public release of the model as it conducts safety tests with a select group of firms, including Goldman Sachs.

Tech pioneer George Hotz dismissed the panic, suggesting that skilled hackers already possess similar capabilities and that the difficulty lies in the incentives for fixing bugs.

However, Stanford's latest AI Index highlights a growing disconnect between AI insiders and a public increasingly anxious about the technology's impact on security.