Seven of Eleven Hotel Groups Let an AI Agent Reach Checkout. Few Chose To

Seven of eleven major hotel groups let at least one AI agent reach checkout in live tests on 18 October 2026, according to research consolidated by Terence Ronson of Pertlink Limited, and most of that openness looks accidental. His Hospitality Net analysis tested Anthropic’s Claude and OpenAI’s ChatGPT against ten hotel groups each.

Agents reached checkout at five or six of each tool’s ten groups. Marriott, IHG, Accor and Choice let both through. Results split elsewhere: Hyatt blocked Claude with a bot challenge while ChatGPT reached the payment page at Andaz Singapore, Hilton let Claude reach payment while ChatGPT hit a 502 error, and neither agent saw a rate at Wyndham. China’s majors Jin Jiang, BTG Homeinns and H World gate booking behind logins, SMS codes or apps, which Ronson treats as channel design rather than anti-agent policy. No bookings were submitted. Gemini reported every group as blocked, a claim he flags as unverified.

The wider study is the telling part. Across 105,002 hotel websites, only 3.3% set any robots.txt rule against AI crawlers. Hotels have not shut agents out; most have not decided anything. Whether a guest’s assistant can book a given property depends on bot-protection defaults, the agent and the route, not on a distribution strategy.

That matters because demand is already moving. Wyndham’s CEO said on the company’s Q2 2026 earnings call that roughly 60% of its guests’ travel searches happen inside an LLM, and the company is positioning its ChatGPT app as a direct-booking play (Hospitality Daily). Ronson reports Wyndham has placed its own agents inside ChatGPT, Claude and Gemini, yet neither test agent saw a rate on its website.

For operators, the practical step is to treat agent access as a decision. Check what your booking engine, bot protection and robots.txt do to an agent today, then choose, route by route, which agents you want through to checkout.