Tech · 12 November 2026 · 4 min read

No Check-In for Bots: United Airlines and eBay Shut the Door on Autonomous AI Agents

In brief: United Airlines has officially blocked personal AI assistants, such as Meta’s Muse and OpenAI’s Dots, from booking flights on its systems, citing non-delegable federal safety rules and a refusal to cede margins to new middlemen. The restriction aligns with recent scraping blocks enforced by eBay and major financial institutions, signaling widespread industry resistance against unauthorized autonomous agents.

by Team Mocchi's

No Check-In for Bots: United Airlines and eBay Shut the Door on Autonomous AI Agents

The grand promise of delegating online purchases to artificial intelligence—allowing bots to track prices, auto-fill passenger data, and execute transactions—is meeting swift and organized corporate resistance. United Airlines has officially barred personal AI assistants from purchasing flight tickets across its digital channels. Concurrently, major online marketplaces like eBay have begun systematically blocking commercial agents such as Meta's Muse and OpenAI's Dots from indexing listings. The vision of frictionless "agentic commerce" championed by frontier labs is running headfirst into digital fortress walls erected by established industry leaders.

The fear of creating a new digital middleman

United’s decision is far from a minor technical patch; it represents a deliberate commercial strategy. Speaking at an industry event covered by The Verge, United CEO Scott Kirby captured the airline's stance succinctly: "To be honest, I don’t want to create another middleman."

Airlines spent two decades waging costly battles against online travel agencies and aggregators to claw back direct bookings and reduce commissions. Handing over the transaction flow to general-purpose agents operated by Meta, OpenAI, or Google poses an existential threat to that hard-won direct relationship. If consumer interactions occur strictly inside conversational interfaces, airlines risk becoming commoditized backend service providers, stripped of profitable upselling opportunities—such as seat upgrades and baggage fees—and severed from their proprietary loyalty ecosystems.

Regulatory compliance: baggage checks cannot be delegated

Alongside economic self-defense, United pointed to strict regulatory mandates that software agents simply cannot fulfill. Under federal aviation regulations in the United States, passengers are legally required to confirm specific disclosures prior to booking, including an explicit declaration that their baggage does not contain hazardous materials or explosives.

An autonomous software agent holds neither the legal capacity nor the factual knowledge to sign such binding attestations on behalf of a human traveler. Permitting unauthorized agents to execute transactions exposes airlines to regulatory fines and severe civil liability in the event of disputed charges, itinerary hallucinations, or compromised customer data. While United will still allow AI assistants to surface schedule and fare queries, the actual booking mechanism must remain exclusively in the hands of authenticated human users on official interfaces.

The retail front follows suit: eBay’s defensive measures

United’s defensive stance is not an isolated incident across the web. As detailed in a report by The Verge, eBay has similarly begun severing access for autonomous shopping agents. Users who relied on Meta Muse to generate daily digests of collector items, such as luxury watches, suddenly found their agents blocked by site defenses or redirected to unverified external mirror sites.

Retail platforms, delivery networks, and financial institutions are reaching the same conclusion: uncontrolled agentic scraping burns server resources, obfuscates fraud detection systems, and eliminates the ad-view impressions that monetize their ecosystems. Until formal business models and verified identities are established, platforms are choosing aggressive firewall rules over open interoperability.

Mocchi's take

This pushback highlights that building effective agentic AI is not merely an engineering puzzle of headless browsing and prompt design, but an intricate challenge of regulatory compliance and commercial alignment. For European and Italian businesses looking to integrate autonomous systems, relying on unapproved scraping or circumventing terms of service is no longer viable. The future of agentic workflows will depend on authenticated API protocols, cryptographic identity verification, and bilateral business agreements, making rigorous data governance the ultimate foundation of any serious enterprise automation initiative.

Further reading

All articles on the Mocchi's blog