Claude Mythos: Early industry responses and implications for cyber risk management

In early April, Anthropic announced Claude Mythos Preview, a frontier AI model with vulnerability discovery capabilities that Anthropic claims surpass those of all but the most skilled human researchers. A public ‘Mythos-class’ model (Claude Fable 5) was subsequently released on 9 June 2026.

However, on 12 June following a US government directive to restrict access to foreign nationals, Anthropic suspended all access to both the Mythos 5 and Fable 5 models. The emergence of Mythos has raised important questions for financial institutions around vulnerability management, remediation capabilities and third-party risk.

Through discussions with our members and the ORX Cyber Community, we're exploring how the industry is responding to these developments. In this blog, we summarise the key discussion points, emerging industry perspectives and common themes identified so far.