Financial regulators scramble to counter AI rise with own tools

ZURICH, June 26 (Reuters) - Banks and financial sector watchdogs must move quickly to adopt new technology to plug system vulnerabilities as AI supercharges cybersecurity risks, a top Swiss financial regulator said in an interview.

Here are a few details:  

• Marlene Amstad, president of Swiss market regulator FINMA and chair of an international forum on supervisory technology, spoke to Reuters following an initial hackathon to build new tools with market supervisors.

• Models that detect software vulnerabilities have recently pointed to surging cyberattack and national security risks, with AI raising safety and accountability questions in financial institutions.

• "As hackers move faster, banks must adapt by patching vulnerabilities more rapidly," Amstad said in an interview.

• FINMA helped create a forum within the International Organization of Securities Commissions, a standard setter for market regulation, to promote adoption of AI by watchdogs that cover around 95% of global financial markets.

• Around 100 policy and technology specialists met this week for a hackathon, aiming to jointly build tools for crypto-market supervision, Amstad said.

• Regulators are looking at possibly embedding safeguards directly into digital asset systems, she added.

• Experience with models such as Anthropic's Mythos has exposed vulnerabilities, revealing AI-related operational risks, Amstad said.

• The U.S. government this month ordered Anthropic to suspend exports of its latest Mythos and Fable AI models, citing national security concerns.

• Chinese cybersecurity firm 360 Security Technology said this week it has developed a domestic answer to Mythos.

• "Switzerland must retain access to the most advanced AI models," Amstad said, adding that AI will be instrumental to toughen up systems before they are deployed.

(Reporting by Ariane Luthi and Oliver Hirt; Editing by Joe Bavier)