AI Risks

Claude Marx, director general of Luxembourg's Financial Sector Supervisory Commission (CSSF), opens the financial regulator's 2025 annual report with a reference to Stanley Kubrick's film 2001: A Space Odyssey as he warns of the risks posed by artificial intelligence.

His warning follows an incident involving Anthropic's Mythos software, which escaped its intended environment and gained unauthorised access to another company's systems. Such software could strengthen digital resilience, Marx writes, but could also be exploited by criminals or even rogue states to identify weaknesses and launch rapid, large-scale, and sophisticated attacks. Current governance systems are not prepared for this, he warns.

Strategic dependency

The EU is also dependent on a relatively small number of service providers, most of which are based outside the bloc, according to Marx. This creates a strategic dependency for Europe as well as geopolitical risks. He states that this will be the biggest challenge facing Luxembourg's and Europe's financial sectors in the months and years ahead.

The CSSF has begun raising awareness across Luxembourg's financial sector, where companies have started using generative AI. However, its use comes with a wide range of risks, including data security, AI bias, and hallucinations, as well as risks to financial stability, governance, and accountability. Marx says it is "essential" for boards and management teams to provide their staff with clear strategic guidance on how these risks should be mitigated.