Program report · filed November 19, 2026
AI in SalesMeasured report
United Airlines Blocks Unauthorized AI Agents From Booking
United Airlines bans unauthorized AI agents from booking or changing reservations on its site and app. Forrester cites IAM gaps, $25-per-site agent hacking and PII risk; four controls recommended.
By Daniel Okafor3 min read558 words
Program notes
- United bans unauthorized AI agents, bots and browser extensions from searching, booking, buying or managing anything on united.com and its app
- United currently works with zero authorized AI agent providers, making the ban near-total
- Amazon, Booking.com, eBay, JPMorgan Chase and Target already restrict AI agents to varying degrees
- Forrester cites AI agents hacking e-commerce sites at $25 per site
- Forrester recommends allowing only verified providers (Anthropic, Meta, OpenAI) with token-based access and step-up human authentication

United Airlines now bars any third-party AI agent, bot or browser extension that searches, books, buys, manages or collects information on united.com or the United app without explicit permission. The airline stated the rule directly: "If a third-party AI agent, bot, browser extension or browser feature searches, books, buys, manages or collects information on united.com or the United app for you, and United hasn't given permission, it's unauthorized."
The policy covers material changes to reservations and ticket purchases. United frames the restriction as both a compliance and product move. "This data helps us meet regulatory requirements while also enabling us to deliver personalized experiences and improvements to our products and services," the company said. "This is a responsibility we take seriously."
United does not yet work with any AI agent providers for authorized access. The approval framework it references has no current participants, which makes the ban near-total in practice.
United is not alone. Amazon, Booking.com, eBay, JPMorgan Chase and Target all maintain varying levels of restrictions on AI agents accessing their sites or customer-facing channels.
Why are B2C sites blocking AI agents now?
A Forrester analysis of the United decision identifies eight drivers, and most concern control rather than customer preference.
- Owning the customer experience. Blocking all agent access long term may prove counterproductive, Forrester notes, but the short-term logic is to keep the relationship direct.
- IAM and fraud tooling gaps. Existing identity and access management and fraud management layers may not detect or manage AI agents at all.
- Avoiding middleman status. Restricting agents keeps organizations from becoming AI intermediaries that merely broker transactions, and prevents agents from flooding sites with queries that degrade service.
- Fraud exposure. Unvetted agents add liability. Forrester cites AI agents hacking e-commerce sites at $25 per site.
- Immature defense tooling. AI agent runtime controls and Know-Your-Agent (KYA) validation are emerging technologies that may not hold up at scale in mission-critical use.
- Regulatory risk. Any loss of personally identifiable information caused by an agent threatens compliance standing and brand reputation.
- Waiting for standards. Until liability schemes, mature agent verification and authentication with human-in-the-loop decisions arrive, most B2C organizations will likely limit third-party agents on commerce sites.
What does this change for commerce teams?
Forrester recommends four controls for B2C sites that want a workable policy rather than a blanket ban:
- Allow only well-known, verified AI agent providers — Forrester names Anthropic, Meta and OpenAI — to act on customers' behalf.
- Mandate short-term, token-based access rights management and delegation, including explicit human-performed step-up authentication.
- Err on the side of caution when customer agents attempt high-risk, high-value transactions — ticket purchases and reservation changes in United's case.
- Audit all agent actions in detail.
The immediate consequence for sellers and travel sellers specifically: any workflow built on a third-party agent touching United inventory is now offside. Longer term, the recommendation implies a shift from blocking to vetted delegation, with token-scoped access and human sign-off at the point of payment. Expect more large B2C operators to follow United's pattern until KYA-class verification and liability frameworks reach production maturity.
Forrester clients can schedule inquiry or guidance sessions on implementing identity and access management for customer AI agents with the analyst.
via forrester.com (Source)
More from Daniel Okafor
Show full bio
Correspondent covering regulation, policy and their impact on Sales technique and methodology, sales enablement at Quota Brief.
127 articles