Policy and Compliance

United Bans AI Agents From Bookings as the EU Moves to Force AI Shopping Agents to Disclose Paid Picks

By Ashit Malik, Socioshoot · Published · Updated · 5 min read · Covers news of 12 November 2026

In short: On 11 November 2026, United Airlines changed its Contract of Carriage to bar AI agents from buying tickets, seats or upgrades on a customer's behalf — agents can still search and show fares, but a human has to complete the purchase, or United can cancel the booking and penalise the account. The same week, the European Commission proposed a Digital Fairness Act that would, among other things, require AI shopping agents to disclose when a product recommendation has been paid for. Different responses, same underlying shift: AI agents are increasingly standing between your business and the customer at the point of purchase, and both platforms and regulators are starting to draw rules around that.

Key takeaways
  • United's new Contract of Carriage rule (6.J.6) bans using AI agents — built-in assistants, AI browsers, or third-party travel tools — to make, modify or service a ticket, whether or not a human is supervising.
  • Agents can still search and display United fares; the actual purchase has to be completed by the customer. Violations let United cancel the booking without notice and penalise the account, up to ending MileagePlus membership.
  • Reported reasons include privacy, unintended purchases, and regulatory pressure. United's CEO has said he wants to avoid "another middleman" between the airline and its customers.
  • The EU's proposed Digital Fairness Act (a 72-page proposal, not yet law) would replace the 2005 Unfair Commercial Practices Directive, with fines up to 6% of worldwide turnover.
  • Article 26 of the proposal would specifically require AI shopping agents to disclose when a suggestion was paid for — and Article 11(6) would make brands and agencies liable for the compliance of influencers they pay, not just the influencers themselves.
  • None of this is law yet — it still needs European Parliament and Council agreement, and would apply 12 months after that. But the direction from both a platform and a regulator, in the same week, points the same way: expect more friction and more disclosure requirements around AI agents transacting on a customer's behalf.

What United actually changed

On 11 November 2026, United Airlines updated its Contract of Carriage with a new rule, 6.J.6, that prohibits using an AI agent to make, modify, or service a booking, or to buy any product or service through United on a customer's behalf. The definition is broad: it covers AI assistants built into consumer platforms, standalone AI browsers, and third-party travel-booking tools — whether or not a human is actively supervising the agent while it acts.

Agents aren't banned outright from the experience: they can still search fares and display results. The purchase itself has to be completed by the customer directly. If a booking is made in violation of the rule, United can cancel it without notice, and penalties range up to ending the customer's MileagePlus membership.

Reported reasoning (via View from the Wing, citing the Wall Street Journal) points to a mix of privacy concerns, worry about unintended or mistaken purchases made by an agent acting too autonomously, and regulatory pressure. United CEO Scott Kirby has reportedly framed it as wanting to avoid "another middleman" standing between the airline and the people actually buying tickets.

What the EU is proposing

In the same window, the European Commission adopted a proposal for a Digital Fairness Act — a 72-page regulation intended to replace the 2005 Unfair Commercial Practices Directive. It's a proposal, not law: it still needs agreement from the European Parliament and the Council, and would apply 12 months after it formally takes effect. A few pieces of it matter directly to advertisers and brands selling into AI-mediated channels:

  • Article 26: would require AI shopping agents to disclose when a recommendation or suggestion has been paid for — the agentic-commerce equivalent of disclosing a sponsored result.
  • Article 11(6): would make brands and agencies liable for the compliance of influencers they pay, not just the influencers themselves.
  • Article 12: would extend the Digital Services Act's existing bans on profiling-based ads for minors, and on targeting using sensitive data, from platforms specifically to all traders — including ad tech generally.
  • Ten specific dark patterns would be banned outright, alongside required subscription-renewal reminders and an online cancellation button.
  • Fines could reach up to 6% of worldwide turnover; coordinated national enforcement actions carry a floor of at least 4% of turnover or €2 million.

Why this matters even though neither is final yet

United's rule is already in effect; the EU proposal is not law. But read together, they're the same story from two different directions: AI agents are increasingly the thing standing between your ads, your site, and the actual purchase — and both a major platform and a major regulator are responding to that by drawing explicit rules around it, in the same week.

If any part of your funnel assumes an AI agent will complete a transaction on a customer's behalf — agentic checkout integrations, AI shopping assistants, affiliate or influencer arrangements where an AI tool surfaces your product — two things are worth tracking as this develops:

  1. Platform-level restrictions on agent checkout may spread beyond travel. If more merchants or platforms follow United's logic, agentic-commerce integrations you're relying on could stop working with no warning beyond a terms update.
  2. Disclosure requirements for paid AI recommendations are coming from the regulatory side regardless of what any single platform decides. If you pay for placement inside an AI shopping agent or a paid-influencer arrangement that an AI tool surfaces, build in the assumption that "this was a paid placement" disclosure becomes a requirement, not a courtesy.

Want help thinking through how agentic commerce and disclosure rules might affect your own funnel? Request a free account review.

New updates by email

Get new marketing updates in your inbox.

We email you when we publish something worth your time. No spam, unsubscribe anytime.

Sources

Facts in this article come from the sources above, checked on 17 November 2026. Platform features roll out in phases, so check your own account before acting. We update this page when official details change.

Ashit Malik founded Socioshoot in 2015. Socioshoot is a Meta Business Partner agency that runs paid media and builds websites for 150+ brands. About the team · Get a free proposal
Keep reading

Related guides.

Policy and Compliance

Italy Fines Meta €805 Million Over Facebook's Photo Gallery Suggestions

Italy's data protection authority, the Garante, fined Meta €805 million after finding that the Facebook Android app read about 20 million Italian users' photos, videos and metadata for a gallery-suggestions feature without specific consent — relying instead on one vague, app-wide permission switched on by default. Meta must now get separate, explicit consent for the feature and keep it off by default, even for users who'd already granted the general system permission. This doesn't change ad targeting rules directly, but it's another reminder that "one broad permission, on by default" is no longer good enough for EU regulators, and the same scrutiny applies to anything advertisers ask customers to opt into.

· 4 min read →
Policy and Compliance

Türkiye's Social Media Age Rules: No Personalised Ads to 15–17s

Türkiye's new social-network regulation took effect on 1 November 2026. It requires age verification for new accounts, bars children under 15 from having accounts and prohibits personalised advertising to users aged 15 to 17. Existing accounts have until 31 March 2027 to complete verification. It is a Türkiye-specific rule, not a global Meta or Google Ads change. Advertisers targeting Turkish audiences should review age breakdowns, youth-focused creative and platform guidance.

· 4 min read →
News Roundup

Performance Marketing News: 12–17 November 2026 Roundup

The main updates between 12 and 17 November 2026: Google Ads added a {search_term_matchtype} parameter that exposes AI Max expansions hiding inside what looks like "exact match" traffic; Italy's Garante fined Meta €805 million over consent for Facebook's photo-gallery suggestions feature; United Airlines banned AI agents from completing bookings while the EU proposed a Digital Fairness Act that would force AI shopping agents to disclose paid recommendations; and Google quietly opened Performance Max household-income exclusions to a beta group of advertisers.

· 4 min read →
FAQ

Common questions.

What did United Airlines ban?

Using an AI agent — built-in assistant, AI browser, or third-party tool — to make, modify or service a booking on a customer's behalf. Agents can still search and display fares; a human must complete the purchase.

When did United's rule take effect?

11 November 2026, via an update to its Contract of Carriage (new rule 6.J.6).

What is the EU's Digital Fairness Act?

A proposed regulation to replace the 2005 Unfair Commercial Practices Directive, including a requirement (Article 26) that AI shopping agents disclose when a recommendation was paid for. It is not law yet.

Is the Digital Fairness Act in effect now?

No. It still needs agreement from the European Parliament and the Council, and would apply 12 months after it formally takes effect.

Why does this matter for advertisers?

AI agents are increasingly involved in the path to purchase. Platforms (like United) and regulators (the EU) are both starting to impose rules on how agents can transact and what they must disclose, which could affect agentic-commerce integrations and paid-placement arrangements.

Free account audit

Tell us your targets. We will tell you what it takes.

Free account audit. We reply within 24 hours, and if we are not the right fit, we will say so.