Resy banned diners whose AI agents grabbed the tables

Share
Resy banned diners whose AI agents grabbed the tables

Reservations became the first consumer service where delegating to your AI assistant is a terms-of-service violation — and the diners getting banned are the ones the platform is quietly building agent access for. Meanwhile, China's biggest AI image company shipped a studio built for the same one-person crew.

Resy has been deactivating the accounts of diners who pointed personal AI agents at its booking system, and it will not say how many. The American Express-owned reservation platform told Restaurant Business it "does not currently permit unapproved third-party bots or agents to independently access or interact with the Resy platform," adding that automated access "can introduce risk to the platform and give some diners an unfair advantage." Resy has banned automated activity for years; what is new is that the enforcement has landed on ordinary cardholders rather than commercial scalpers, and that the company declines to put a number on it.

The cases that went public are specific. JC Bahr-de Stefano, a New York venture capitalist, set the AI assistant Instinct to watch 4 Charles Prime Rib in the West Village and posted the deactivation email himself — he says the agent made roughly 200 API requests an hour around the clock, with bursts of 200 to 375 each morning when the restaurant's books opened. Brian Distelburger, cofounder of the performance-review startup Windmill and a 20-year American Express Platinum holder, gave his own agent his Resy credentials and was cut off within about 48 hours. "It never booked a single reservation," he told Business Insider. He was reinstated only after that same agent drafted a multi-paragraph appeal, which Resy granted as a "one-time courtesy."

Why it matters: the platform's position is more defensible than the enforcement looks. Resy cofounder Ben Leventhal has said on the record that "10,000 vibe coders have written Resy sniper bots," and the failure mode is real — one Los Angeles investor's agent booked an unrequested Tokyo restaurant with a full cancellation policy and ate a roughly $200 fee. But Resy also ships booking integrations with ChatGPT and Claude, and its statement promises it will "continue to evaluate opportunities thoughtfully." So the rule is not "no agents." It is "no agents we did not approve," enforced retroactively on users who had no way to know which side of that line their assistant was on. Restaurateurs are reading it differently: Dhamaka co-owner Roni Mazumdar called it "dehumanizing the last standing experience that is still totally human," and Jean-Georges chief operating officer Jeremy Mustakas framed the line as "the technology should open doors, but it shouldn't create a digital velvet rope."


LibTV 1.5 turns one person into a full animation crew, and its maker is worth more than $2 billion. The Chinese platform from Evoken — the company formerly known as LiblibAI, which raised a $300 million-plus round in June at a post-money valuation above $2 billion — shipped five new features in its 1.5 release, per AI Era: a character studio with more than 100 face and wardrobe adjustments that binds a voice to a saved digital actor, a screenwriting module that drafts a script framework in about ten minutes across eight genres, a director mode that breaks a rough idea into shots, a natural-language 3D previz studio that auto-builds sets and camera moves, and one-click title sequences. A hands-on reviewer built a full Chang'an city scene in the previz tool in roughly five minutes for about 330 credits.

The pitch is production time, not model quality — AI Era's headline claim is finishing an episode in an afternoon against three all-nighters before. That is an editorial anecdote, not a benchmark, but the direction is what matters: the tooling is consolidating script, storyboard, shots and edit into one canvas, and the output is aimed at the comic-drama format that has become China's highest-volume AI video category. We covered how far that format has already gone — China's prime-time TV slot now belongs to an all-AI drama.

What to watch: whether Resy publishes an approved-agent programme or keeps the rule opaque, and whether anyone builds the agent-to-agent reservation market Leventhal is clearly bracing for.

Should platforms be required to say which agents they allow before they ban you for using one? Tell us in the comments.

Read more

OpenAI drops 372 math results, nearly all from one prompt

OpenAI drops 372 math results, nearly all from one prompt

A dense news day across three fronts: OpenAI published a mountain of mathematics from an unreleased model, SpaceX reportedly wants the debt market to fund its Nvidia orders, and Anthropic restructured its cyber program around three tiers of trust. OpenAI published 372 mathematical result families from an internal frontier model — and says nearly all of them came from a single prompt handed to a single AI agent. The release comprises 722 manuscripts touching hundreds of open questions, plus Lea

Underdog launches a private on-device AI assistant, backed by a16z

Underdog launches a private on-device AI assistant, backed by a16z

The privacy split in consumer AI got a new entrant tonight, GitHub turned code-review benchmarks into a vendor-bias argument, and OpenAI opened its oddest API to everyone. Underdog launched in invite-only beta: an on-device AI assistant that keeps your data on your machine and never charges you a subscription. Self-taught coder and Thiel Fellow Sigil Wen — who moved to Silicon Valley at 17 and lived in an AI hacker house with Andrej Karpathy — built his own inference engine, Husky, to run a 2

Today in AI — October 6, 2026

Today in AI — October 6, 2026

A day of second-order moves: the labs are buying task data from software vendors instead of scraping the web, Waymo is borrowing to fund the robot world, and the biggest bank in the US just put a number on what Anthropic's latest model cost it in risk. Models & Research * OpenAI is training GPT-6 Astra on Ironclad's real contracting work. Ironclad staff helped turn 11 tasks across legal, commercial and procurement work — setting up NDAs, approval workflows, clauses that change by jurisdict

Meta, Walmart and Stripe publish the Personal Agent Protocol

Meta, Walmart and Stripe publish the Personal Agent Protocol

The agent economy is writing its rulebook tonight: one open standard for AI bots at the checkout, a cheaper image model from Google, and Anthropic turning bug-hunting into a tiered product. Meta, Walmart, Stripe and Sierra are publishing an open "personal agent protocol" — a standard that defines how personal AI agents interact with businesses online. The group behind it reads like a cross-section of agentic commerce: Meta, Sierra, Walmart, Stripe, Shopify, Genesys, Rocket, NiCE, Decagon and I