JPMorgan Chase has announced a policy to block external AI agents from accessing customer accounts. The bank disclosed this decision during its third-quarter earnings call on October 13, 2026, citing concerns over security, client protection, and the potential for software to act without sufficient oversight.
CEO Jamie Dimon highlighted the risks associated with rogue AI agents and the ambiguity regarding liability if an unauthorized tool performs unintended financial transactions. CFO Jeremy Barnum confirmed that the bank is actively working to prevent third-party access, specifically noting risks related to credential sharing and screen scraping.
The bank is currently developing its own internal agentic AI tools capable of executing multi-step tasks. This follows a previous restriction in June 2026, when JPMorgan limited its Hong Kong staff's access to Anthropic's Claude due to compliance and licensing concerns.
While the bank aims to maintain control over customer data and relationships, executives acknowledged that completely blocking all external agents remains a technical challenge. The institution intends to prioritize its proprietary AI developments to provide automated services while maintaining strict security standards.
