Particle.news
Download on the App Store

Bank of Italy Engages AI Firms Over Mythos Cybersecurity Risks

The central bank has opened talks to shape how powerful new models are rolled out so banks, vendors and regulators can ensure safe governance and resilience.

Bank of Italy Governor Fabio Panetta attends the G7 Finance Ministers and Central Bank Governors' Meeting in Stresa, Italy May 24, 2024. REUTERS/Massimo Pinca/File Photo
Governor of the Bank of Italy Fabio Panetta presents the Bank of Italy's annual report, in Rome, Italy May 30, 2025. REUTERS/Remo Casilli

Overview

  • The Bank of Italy has begun direct contacts with global AI providers and convened meetings with national authorities, banks and IT vendors to assess risks from new models before wider use.
  • Supervisors have focused scrutiny on Anthropic's Mythos, a model built to find software flaws that security teams can fix but that experts warn could be repurposed to speed up cyberattacks.
  • Governor Fabio Panetta said banks remain responsible for protecting systems and that the same duties apply to third‑party tech firms when services are outsourced.
  • Panetta urged executives to fund stronger defences with part of recent record profits and to set executive-level governance, clear responsibilities and rapid-response plans rather than relying on technology alone.
  • The regulator's consultative stance signals a wider European push for vendor engagement, shared third‑party accountability and targeted bank investment, which could change how lenders vet and adopt advanced AI tools.