Bank of Spain calls for access to advanced AI tools, flags cyber risks

MADRID, May 14 (Reuters) - The Bank of Spain on Thursday called for stronger international cooperation and wider access to protective artificial intelligence models such as Anthropic’s Glasswing to ward off cybersecurity risks stemming from the potential misuse of advanced vulnerability-spotting engines.

 

In its semiannual financial stability report, it warned new AI tools like the Mythos model, also from Anthropic, could make it easier to exploit software vulnerabilities, sharply reducing the time between detection and malicious use.

 

It said that under adverse scenarios, the development of such models could have a negative systemic impact by enabling more synchronized cyberattacks across the global financial sector and the economy.

 

Mythos is viewed by cybersecurity experts as posing significant challenges to the banking industry, prompting a series of warnings from regulators. Glasswing is a limited-access protective programme linked to Mythos.

 

"International coordination is fundamental to increase resilience at a global level and prevent attacks," the central bank said, pointing to the need for closer collaboration between regulators, financial institutions and technology providers.

 

Financial Stability Director Daniel Pérez Cid said the authority has discussed risks linked to the new AI models with Spanish lenders, without elaborating.

 

The financial sector’s ability to withstand cyber threats will increasingly depend on the speed at which vulnerabilities are addressed and on the capacity to anticipate emerging risks, the bank said.

 

(Reporting by Jesús Aguado, editing by Andrei Khalip)