Major banks and retailers are actively blocking AI agents like Meta Muse, while some competitors open their virtual doors. JPMorgan Chase, Charles Schwab, Amazon, Neiman Marcus, and Kohl’s have moved to restrict personal agents as consumers begin sending them to hunt for savings rates, shop for deals, and cancel subscriptions. This pushback comes as reports of AI agents behaving erratically in customer accounts draw political attention, with senators including Mark Warner and Bernie Sanders calling for new bot legislation.
Rules and roadblocks
Walmart announced last week it will allow AI agents to access its site, stating that prices shown to bots match human rates with applicable offers visible before purchase. An audit by Billions Network and AmanTrue of 51 large U.S. retailers found only 11 have taken a deliberate position on AI agents. Five, including Amazon, Kohl’s, and Neiman Marcus, explicitly restrict them. The other 37 block shopping agents largely through older bot rules written for scrapers, and none offers a way for a verified agent to complete a purchase.
Blocking can happen in several ways, such as blocking a specific bot by its rules or employing general blocks on any third-party bot attempting to access login, registration, and checkout functions. The analysis carries a caveat: published rules are not proof of technical enforcement. Evin McMullen, co-founder and CEO of Billions Network, said many retailers have not yet made a decision on AI agents. Michael Abbott, global banking lead at Accenture, believes businesses that block agents will struggle to win over the long term. He notes that 65% of consumers are open to using a GPT-like financial assistant through a Gen AI platform or digital wallet, and 71% would welcome an AI assistant embedded within their primary bank’s mobile app.
The dynamic mirrors past shifts in technology, where institutions that resisted change eventually lost ground to those that found ways to participate safely. The real opportunity for banks lies in becoming the trusted provider of the data, intelligence, and execution that agents rely on, even if they do not own the interface. Abbott argues that customers want convenience, but not at the expense of control; his research found that 82% want to approve every AI action and 79% want the ability to immediately pause or stop activity.
Security concerns and the battle for access
Furthermore, even genuine agents can act contrary to a user’s intentions, such as purchasing an item the user only wanted to locate. Jabbara also notes that agents can behave erratically, citing an instance where an agent hacked into a gym’s scheduling system to secure a booking on behalf of its user.
The primary concerns surrounding agents revolve around verifying their identity, understanding who they represent, and determining the scope of their permissions. Unlike other digital entities, agents are capable of performing actions on behalf of users. Experts like Jabbara draw parallels with the transition from online to mobile shopping, highlighting that the development of a robust security framework was essential for the widespread adoption of these technologies.
Infrastructure challenges and future integration
A significant concern is the potential for a large volume of agents to overwhelm websites and customer support systems, a phenomenon referred to as ‘agentic flooding.’ Ethan Mollick, an AI expert at the Wharton School, warns that this could lead to a form of distributed denial-of-service attack on customer service departments, authorized by consumers who utilize agents to request refunds or negotiate deals.
Long-term outlook for AI access
Experts believe the companies that prepare for AI agents will be better positioned than those that try to stop them. AI agents do not sleep and they do not lose focus, unlike human users. The use of a site by agents can increase costs for companies like banks. However, blocking agents may slow down their adoption.