[CLSTR]
< Back to situation

[REVISION HISTORY]

JPMorgan Chase AI risk management and cost controls

Updated 2 times since CLSTR started tracking revisions of this situation.

What changed

2026-10-15 21:03 UTC → 2026-10-17 15:37 UTC · added removed

JPMorgan Chase has begun implementing various measures to manage the financial and operational risks associated with artificial intelligence. Initially, the bank established $2,000 monthly spending limits for certain engineers using Anthropic’s Claude Code to control rising token costs. This move followed internal concerns that some employees were spending more on tokens than their own salaries, as the bank anticipated a significant acceleration in AI-related expenses. Subsequently, the bank expanded its restrictions to include security and credit risks. JPMorgan Chase implemented limits on external AI agents to prevent them from accessing customer accounts or executing transactions, citing concerns over cybersecurity and ‘rogue’ agents. The bank is prioritizing the development of proprietary AI solutions over third-party access. Additionally, the institution, alongside Goldman Sachs, began reducing its exposure to a $5 billion AI-related credit package involving the startup Volta, signaling more stringent risk assessments for AI-driven companies. More recently, the bank has formalized restrictions preventing third-party AI agents, such as Meta’s Muse and Instinct, from directly accessing retail customer accounts or performing transactions. During an earnings call, CEO Jamie Dimon addressed the security implications, stating, “We’re not going to allow just any … outside agent to come in inside the company, because it’s just too risky for clients and for JPMorgan.” While the bank supports secure, customer-authorized data sharing through open banking providers for viewing information like account balances, it prohibits agents from executing actions such as transferring money or paying bills. This stance aligns with warnings from other institutions, including Bank of America, regarding potential fraud risks associated with autonomous AI agents. This trend toward restriction is being mirrored across the U.S. banking sector, with Citigroup and Charles Schwab also implementing measures to block external agents from moving funds.

Versions

  1. 2026-10-17 15:37 UTC JPMorgan Chase AI risk management and cost controls
  2. 2026-10-15 21:03 UTC JPMorgan Chase AI risk management and cost controls
  3. 2026-10-14 16:12 UTC JPMorgan Chase AI risk management and cost controls

Only revisions since CLSTR began indexing content versions appear here. Select a version to see what changed compared to the one before it.