European Union officials are increasing pressure on Anthropic to provide access to its unreleased AI model Mythos, as concerns rise among regulators that the technology may expose vulnerabilities across the region’s banking system.
On Wednesday, 6 May, officials from the European Commission and the EU’s cyber agency ENISA will scrutinize the risks of the Mythos model in the European Parliament.
However, Anthropic will not be in attendance due to the “short notice” of the invitation, which Dutch European Parliament lawmaker Kim van Sparrentak called “extremely worrying”.
Mythos, an advanced model designed to identify flaws in computer code, has attracted significant attention from governments and financial institutions after Anthropic restricted access to a small group of mainly US-based companies over cybersecurity concerns.
Experts have warned that the system could be capable of both strengthening software security and accelerating cyberattacks if misused.
European Economic Commissioner Valdis Dombrovskis confirmed on Monday night that the European Commission has already held discussions with Anthropic regarding the tool’s technical capabilities and associated risks.
“The commission representatives met with Anthropic and were briefed on technical details around cyber capabilities and the risk of this Mythos preview,” Dombrovskis said, adding that officials are assessing the implications “in light of EU policies and legislation.”
Brussels is understood to be seeking a pathway for European banks and companies to undergo cybersecurity resilience testing using the model, amid concerns that institutions could be left exposed if US firms gain earlier visibility into vulnerabilities identified by Mythos.