Realtime AI News
United Airlines Becomes First U.S. Carrier to Block Third-Party AI Agents From Booking Flights
United Airlines said Wednesday it will restrict unauthorized third-party personal AI agents from booking travel on its own platforms and through online travel agencies and aggregators. The carrier is the first major U.S. airline to state such a policy explicitly, citing credential sharing, unintended purchases and possible conflicts with FAA and DOT rules.
United Airlines moved on Wednesday to shut personal AI assistants out of its booking flow, becoming the first major U.S. carrier to say explicitly that third-party AI agents cannot act on its flights. The restriction covers United's own digital platforms and, indirectly, access through online travel agencies and aggregators.
"United is taking steps to restrict access — directly within our digital platforms and indirectly via travel agencies and aggregators — to prohibit use by unauthorized, third-party personal AI agents or assistants," the airline said in a statement reported by Skift.
The carrier's stated reasons are practical rather than philosophical. United said handing account credentials to a third-party agent can expose sensitive passenger information, and warned that automated tools can trigger unintended purchases, account changes or MileagePlus redemptions that the traveler never meant to authorize.
United also pointed to regulation. It argued automated AI purchases could run into Federal Aviation Administration and Department of Transportation rules, singling out check-in, where a person must confirm their bags contain no banned substances such as explosives. An agent clicking through that attestation on a traveler's behalf sits awkwardly with the requirement.
The policy lands in the middle of a consumer AI booking wave. Meta CEO Mark Zuckerberg has called travel a top use case for the company's Muse agent, OpenAI acquired AI travel agent Odessia on Tuesday, and Southwest has launched a ChatGPT plug-in for flight search.
Airline leaders are uneasy about the shift. Delta CEO Ed Bastian has said he is "cautious" about AI agents handling bookings, while United CEO Scott Kirby told an event in Chicago, according to The Wall Street Journal, "I don't want to create another middleman."
United said it will review the policy as personal assistants evolve and is updating its Digital Trust Center to explain the restrictions. For agent developers, the practical question has shifted to distribution: a travel agent that can plan a trip but cannot complete a booking on a major U.S. carrier is a partial product.
What to watch next is whether other carriers follow. Skift framed United as first rather than alone, and the same arguments about credentials, liability and DOT compliance apply across the industry. If Delta, American and others adopt similar language, agent builders will face a patchwork of per-airline permissions instead of a single consumer channel.
Why it matters
United turns a vague worry about agents into a concrete liability question: booking a flight involves credentials, consumer-protection rules and a regulatory attestation that someone must personally make. If rivals copy the policy, AI travel assistants will need permission carrier by carrier rather than one integration.
Nearby Updates
All11/12, 05:56
Alibaba opens Qwen4-Max as a 15-day preview with a 1M-token context window
Alibaba opened Qwen4-Max as a fifteen-day preview on November 11, available only through its Token Plan and retired on November 25. The multimodal flagship offers a 1 million token context window with text, image and video input, but ships without published parameters, weights or benchmark results.
11/12, 06:52
Trellis Data unveils Agentaus 1, a 220B-parameter Australian model pitched as cheaper and locally controlled
On Thursday, Canberra-based Trellis Data unveils Agentaus 1, a roughly 220-billion-parameter model about 50 percent larger than its predecessor, built by a company with under $10 million in external funding. Founder Michael Gately concedes it ranks only 43rd globally and trails frontier models by four to six months, but argues a local model is cheaper and safer on data privacy for businesses that need good-enough AI.
11/12, 06:52
Claude Mythos moves into Cisco and IBM security platforms
Anthropic's Claude Mythos is expanding into Cisco's and IBM's security platforms, according to a Security Boulevard report. The signal is less about model capability than about distribution: the model lands inside the platforms security teams already run instead of arriving as a separate endpoint.
11/12, 04:33
Anthropic IPO filing discloses $1 billion Microsoft stake as Microsoft shares slip
Anthropic's IPO filing has disclosed a Microsoft stake worth about $1 billion, and Microsoft shares fell after the report on November 11, according to SSBCrack. The disclosure puts the two companies' mix of partnership and competition in front of public-market investors for the first time.