Anthropic probes unauthorised access to Mythos model

Anthropic says it is investigating unauthorised access to its Mythos Artificial Intelligence model, according to French news agency AFP. Mythos is intended to identify security holes that have existed for decades but now appears to be vulnerable itself. Earlier this month, Anthropic made a restricted release of Mythos to 40 major technology firms, to give them a head start in fixing cybersecurity vulnerabilities before they could be exploited by attackers.