Banks are warning customers they could be liable for an AI agent's mistakes Ally Bank, Chase Bank and Citi have warned customers in emails, app notifications and website banners that they may be held liable for mistakes made by AI agents acting on their accounts, Business Insider reported. Ally told customers earlier this month that "when you use an AI service, you're responsible for any actions that the AI takes," while Chase said in a November email that "authorized transactions may not be reversible if your information is misused" and Citi cautioned that "AI assistants can make mistakes, and that means you can be liable for errors it makes." The warnings arrive as agents such as Meta's Muse, Instinct AI and OpenAI's Dots execute purchases and bookings, prompting Target to deem AI-agent shopping "considered transactions authorized by you" in March and Amazon to block Muse agents in September; JPMorgan Chase CEO Jamie Dimon said on an October earnings call, "we're not going to allow just any outside agent to come in inside the company because that's just too risky for clients and for JPMorgan. Banks are warning customers they could be liable for an AI agent's mistakes Business Insider https://www.businessinsider.com Several banks have recently warned customers that handing over financial power to AI agents comes with some big risks. - Did your AI agent https://www.machinebrief.com/glossary/ai-agent send money to the wrong person? Your bank has some bad news for you… - Banks have recently warned customers may be liable for AI agent mistakes. - It comes as customers continue to hand over financial power to AI agents. Did your AI make a bad financial decision? Well, don't be surprised if you're left holding the bag. That's the warning that several major banks have recently sent to their customers. In emails and app notifications reviewed by Business Insider, banks are essentially telling account holders: AI agents with access to financial information may be treated as an extension of the user — not as an outside intruder. "When you use an AI service, you're responsible for any actions that the AI takes," Ally Bank emailed customers earlier this month. "AI can make mistakes, hallucinate, misunderstand your instructions, or follow them in ways you didn't intend, which could lead to significant loss." In a November email, Chase Bank also warned customers that "authorized transactions may not be reversible if your information is misused." Citi added a banner to its website and mobile app directing users to a page that cautioned: "AI assistants can make mistakes, and that means you can be liable for errors it makes." The fresh alerts are rolling in as a new class of AI agents develops a shopping habit. Previously, AI models like OpenAI https://www.machinebrief.com/glossary/openai 's ChatGPT https://www.machinebrief.com/compare/chatgpt-vs-claude or Anthropic https://www.machinebrief.com/glossary/anthropic 's Claude https://www.machinebrief.com/glossary/claude answered user questions with blocks of text. Now, online agents like Meta's Muse, Instinct AI, or OpenAI's Dots are capable of going online to finish tasks on a customer's behalf — like booking a flight, reserving a restaurant table, or buying deodorant before you run out. Retailers have had to respond to the tools by changing their Ts and Cs. In March, Business Insider reported that Target had changed its customer agreement https://www.businessinsider.com/targets-google-gemini-ai-shop-terms-update-2026-3 to state that AI-agent shopping was "considered transactions authorized by you." In September, Amazon blocked Muse agents https://www.businessinsider.com/amazon-blocks-meta-muse-ai-agent-shopping-site-2026-9 from making purchases, citing safety and reliability concerns. Banks are also trying to draw a line between authorized account connections with AI tools and customers simply handing an AI tool their login credentials. Last month, Axios https://www.axios.com/2026/10/22/ai-agents-banks-unauthorized-accounts-access reported that major banks had detected third-party agents operating in customer accounts, in some cases with customers' permission and login credentials. Every bank email viewed by Business Insider warned customers not to share important information, like usernames, passwords, or security codes. Wells Fargo also said customers should not hand over access lightly. "No technology should be trusted to act on a customer's behalf without strong security, clear customer authorization, and safeguards designed to protect their money and financial information," a representative for the bank told Business Insider. During an October earnings call, Jamie Dimon, the CEO of JPMorgan https://www.businessinsider.com/jamie-dimon-jpmorgan-warning-ai-agents-into-bank-account-2026-10 Chase, issued a similar warning to customers about the use of AI agents. "If people are using them, they should understand the risk," he said. "In the meantime, we're not going to allow just any outside agent to come in inside the company because that's just too risky for clients and for JPMorgan." The awareness campaign has extended to the device and software-makers themselves. The beta of Apple's latest Mac software includes warning language https://www.macrumors.com/2026/11/16/macos-27-3-full-desk-access/ when a user is considering granting an AI agent or app system-level access to their computer. Business Insider https://www.businessinsider.com/bank-liable-ai-agents-mistakes-2026-11 Get AI news in your inbox Daily digest of what matters in AI. Key Terms Explained AI Agent https://www.machinebrief.com/glossary/ai-agent An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve goals. Anthropic https://www.machinebrief.com/glossary/anthropic An AI safety company founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei. Claude https://www.machinebrief.com/glossary/claude Anthropic's family of AI assistants, including Claude Haiku, Sonnet, and Opus. OpenAI https://www.machinebrief.com/glossary/openai The AI company behind ChatGPT, GPT-4, DALL-E, and Whisper.