Bundesbank chief: Anthropic must provide equal access to the Mythos model

The President of the German Federal Bank (Bundesbank), Joachim Nagel, has called on Anthropic to provide equal access to the Mythos model to all organizations directly affected by it. According to him, only by adhering to the principle of a level playing field — equal conditions for all stakeholders — is it possible to conduct an objective and independent assessment of both the model's potential and its risks. The position of the Bundesbank chief is not merely a political statement.

It reflects deep concern among European financial regulators about how rapidly advanced AI systems are penetrating critical sectors without proper supervisory mechanisms. If regulators, central banks, and affected organizations do not have access to the model on equal terms with major commercial clients, an information asymmetry emerges — and this is precisely what troubles the German banker first and foremost. Mythos is one of Anthropic's most powerful models, introduced in 2026.

The company positions it as an advanced reasoning system capable of solving complex professional tasks: financial analysis, legal expertise, scientific research, and strategic organizational planning. This application profile directly aligns with the tasks of financial institutions — and this is precisely why Nagel's request appears systemic rather than incidental. The European context is crucial here.

The EU Artificial Intelligence Act (AI Act) requires developers of high-risk systems to undergo independent assessment and provide technical documentation. For models with the highest potential impact — the so-called GPAI models with systemic risks — the requirements are even stricter. However, a mechanism for direct regulator access to models themselves for independent testing has not yet been standardized.

Nagel raises precisely this question: compliance on paper is one thing, the ability to independently study the system is quite another. Central banks have found themselves in a dual role. On one hand, they themselves research the application of AI in their own work: forecasting macroeconomic indicators, monitoring systemic risks, and real-time financial stability monitoring.

The European Central Bank and a number of national regulators have already launched pilot projects on using large language models in internal processes. On the other hand, these same institutions are obligated to monitor how commercial banks and financial companies deploy external AI systems in lending, risk management, and automated trading. Both directions presume direct access to the model, not merely analysis of documentation from the developer.

Anthropic has not yet commented on Nagel's statement. The company traditionally emphasizes responsible development: Constitutional AI, red teaming with external experts, Responsible Scaling Policy — all of this is part of its public positioning in the area of safety. However, declared openness and provision of technical access to regulatory bodies are fundamentally different things.

The practice of controlled regulatory access to commercial AI models has not yet become an industry standard. The statement of the Bundesbank president is a signal that regulatory pressure on leading AI developers in Europe will continue to intensify. The gap between the speed of bringing new models to market and the possibilities for their independent assessment is becoming a structural problem.

When institutions such as the Bundesbank publicly set the condition of equal access, it creates a precedent — and strengthens the position of those advocating for stricter regulation of AI in Europe's financial sector.