BNP Paribas has taken the significant step of restricting employee access to Anthropic’s Claude AI models for its Asia-based staff. This development is a reflection of the growing trend among major financial institutions in the region to tighten governance around third-party generative AI tools. By putting these controls in place, BNP Paribas joins a league of banks that are increasingly vigilant about the use of advanced AI technologies within their operations.
In a similar move, Goldman Sachs enacted a location-specific restriction for its Hong Kong employees regarding access to Claude around April 2026. The bank pointed to important issues such as contractual compliance and data security as reasons for this precautionary stance.
Following global developments, Anthropic decided to suspend access to its most sophisticated models, including Claude Fable 5 and Mythos 5, from June 13, 2026. This decision was influenced by a U.S. government export control order that seeks to place significant restrictions on AI technology access, particularly for foreign nationals. As a result, this has added further complexity for financial institutions operating in Asia, as they now face the challenge of ensuring that their use of AI technology does not infringe upon U.S. trade regulations.
The implications of this export control order are profound, as banks now find themselves navigating compliance risks while striving to maintain competitive advantages. This situation underscores a crucial point: financial institutions must closely monitor how their employees interact with AI tools and be acutely aware of potential trade violations that may stem from unauthorized access to certain AI models.
In a proactive response, BNP Paribas launched its own LLM-as-a-Service platform in June 2025, aimed at enhancing secure AI deployment within its infrastructure. This initiative allows the bank to provide AI capabilities to its employees securely while safeguarding sensitive financial data from exposure to third-party systems. By leveraging in-house resources, BNP Paribas aims to reduce dependencies and mitigate risks associated with external generative AI tools.