Anthropic's Mythos AI Risks

The rollout of Anthropic's Claude Mythos model has triggered a high-level debate between tech innovators and security experts over the safety of advanced AI. While the U.S. government is negotiating access to the tool for federal agencies, banking officials warn that the model could be used by hackers to sabotage financial networks.

Anthropic has defended its responsible approach, emphasizing that the model was withheld from the general public precisely due to its potent bug-finding capabilities.

However, security researchers have already identified what they describe as 'design flaws' in the model's protocol that could put hundreds of thousands of servers at risk of takeover.