Anthropic CEO Says Chinese Open-Source AI Models Could Catch Up With Mythos Within Six Months

Anthropic's most powerful model, Mythos, can discover zero-day vulnerabilities and detect cybersecurity threats, potentially accelerating both offensive and defensive operations while creating new risks for financial infrastructure. Because open-source models can spread and be replicated quickly, the White House and five major banks have begun urgently assessing the potential impact to guard against the large-scale exploitation of vulnerabilities.

As of July 19, 2026, Anthropic CEO Dario Amodei warned that Chinese open-source AI models could match Mythos's cybersecurity detection capabilities in as little as six months and within about 12 months at the latest. He urged governments and financial institutions to use this window to prioritize patching vulnerabilities in critical systems and strengthening their defenses.