US banks and major retailers block AI agents from customer accounts

By Splitfeed EditorialLast updated Oct 19, 2026, 12:10 UTC

Between October 17 and 19, 2026, a wave of restrictions on AI agent access became public across several major American companies. Amazon blocked Grok's shopping agent, eBay moved to shut out similar tools, and Delta confirmed blocks on AI agents attempting to book flights. On the financial side, JPMorgan, Charles Schwab, Citi, and Bank of America all restricted or outright blocked AI agents from accessing customer accounts. Yelp placed its data behind a paywall specifically to limit agent access. Meta's Muse was among the AI agents named as affected by these measures. The timing and breadth of the moves, spanning retail, travel, dining, and banking within a 48-hour window, suggest coordinated awareness of the issue across industries rather than isolated individual decisions.

The significance here goes beyond any single company's terms of service update. AI agents, software tools designed to act autonomously on a user's behalf, represent one of the more commercially ambitious applications of recent AI development. Their promise is that a user can delegate tasks like booking a flight or rebalancing a portfolio to a program that will execute those tasks without further human input. The companies now blocking these agents control the platforms and accounts those agents need to function. That puts them in a structural position to limit the entire category, regardless of what AI developers build. What is genuinely in dispute is whether these restrictions protect consumers, protect corporate data and revenue models, or both.

The Anti-AI camp has pointed to the specifics of these blocks with apparent approval. The account Matt needs a UX gig paraphrased the scope clearly, noting that Delta, JPMorgan Chase, Charles Schwab, Yelp, and Amazon all moved against agents in the same week, with Yelp paywalling its data and Amazon blocking Grok's shopping functionality specifically. The framing treats these restrictions as reasonable pushback against autonomous systems acting inside sensitive accounts. The Pro-AI camp has not yet published reactions to this story, though that camp would typically argue that blocking AI agents limits consumer autonomy and slows the development of genuinely useful automation tools. The Middle Ground camp has also not weighed in publicly, but would be expected to raise questions about whether the issue is the agents themselves or the absence of agreed-upon standards for how they authenticate and operate.

The argument is likely to sharpen around whether any of these companies, or industry bodies, move to define formal access frameworks for AI agents rather than simply blocking them outright. Regulatory attention to account access rules, or the emergence of an authentication standard that satisfies both platform security concerns and AI developers, would be the clearest signal of which direction this resolves.