Experts Push Back on Mythos 'Hysteria': Say Older AI Models Could Already Find Same Vulnerabilities

CNBC·medium signal

CNBC reports that cybersecurity experts and AI researchers are pushing back on the alarm around Anthropic's Mythos, arguing that the vulnerability-finding capabilities are 'achievable using older models, including those from Anthropic and OpenAI.' The debate centers on whether Mythos represents a genuine step-change in offensive AI capability or if Anthropic is overstating the novelty. The counterpoint matters for builders assessing actual risk.