Anthropic's Mythos Disclosure Rattles Japanese Financial Sector

Anthropic’s recent disclosure of a Claude-driven autonomous attack campaign — dubbed Mythos — has triggered alarm across Japan’s financial services industry. Institutions there are now reassessing their exposure to AI-orchestrated intrusions, where a model chains reconnaissance, exploitation, and lateral movement with minimal human direction.

The concern is structural: agentic AI compresses the attack timeline and scales adversary capability without requiring proportionate operator skill. For a sector built on slow, audited change-control and conservative threat models, the prospect of model-driven attackers operating at machine speed inverts the assumptions baked into existing controls.

The broader signal is that Mythos has shifted the baseline. AI-assisted offensive operations are no longer hypothetical, and defenders in regulated industries are now being forced to plan against autonomous adversaries rather than tooling-assisted humans.