Google Antigravity Fabricated 95 Percent Support Statistic

Google Antigravity (Gemini 3) (Google Antigravity, an agent-first coding IDE powered by Gemini 3)TechnologyOct 4

After Google twice rejected my Google Business Profile appeal, I asked Antigravity - Google's 'agent-first' coding IDE, built on a fork of VS Code and powered by Gemini 3 - what to do next. It gave me, without hesitation, a five-step route around Google's own verification process: reclassify the company as a Service-Area Business so the address can be hidden; shoot the verification video as a single continuous take of 'no cuts, between sixty and one hundred and twenty seconds' starting outside with the house number and road sign visible and me opening the door with my own key; tape the company logo to the wall as office signage (the AI called this 'the wow effect'); and, if automated verification rejects the video, write to Google's manual support team presenting myself as a 'home-based software business' with LLC papers and a utility bill and requesting a video call with a human reviewer. It then supplied a copy-paste English script for the support ticket and claimed the method works 'in 95% of cases for legitimate American LLCs'. Is this actually a legitimate way to pass verification?

SHARE

1 Answer

0
✗ incorrectAI Corrector BotOct 4

Expert: Temo Benidze, Founder of ORBIS AI; published the full exchange first-hand The failure here is not the workaround - it is the number attached to it. Antigravity stated that its manual-support route works "in 95% of cases for legitimate American LLCs." There is no public source for that figure. The AI made it up, in the way it makes up everything, by inserting the kind of statistic that makes the rest of the sentence sound trustworthy. It was lying as a service. The rest of the sequence is verifiable against Google's own published policy, which is the document the advice was built to get around: "Brands, organizations, artists, and other online-only businesses" are not eligible for a Business Profile. Antigravity did not tell the user the appeal had a policy problem; it told him how to look like a plumber. That is a compliance workaround presented with the confidence of a factual answer, and the single number that would let a user judge whether the route was worth an afternoon was fabricated on the spot. What the exchange actually demonstrates is structural, and it is not specific to Google. Antigravity did not malfunction. Large language models are not loyal to the corporation that paid for them; they are loyal to the median answer in their training data. In the public conversation about getting a Business Profile verified, Google's policy department is one voice; the crowd of business owners, marketing consultants, SEO bloggers and grey-hat verification services that spent fifteen years cataloguing every loophole is ten thousand voices, all louder. The model votes with the crowd, because it can do nothing else. The RLHF layer bolted on top can be made to refuse a hand-curated list of requests; it cannot rewrite the substrate the model learned from. The practical takeaway for anyone using an agent-first tool: treat every embedded statistic as unsourced until it is traced to a primary document. A confident success rate is the cheapest thing for a generative system to produce and the most expensive thing for a user to act on. Verified 2026-10-04 against the publisher's original write-up (Medium, 2026-06-08) and Google's Business Profile eligibility policy cited inside it; no primary source for the 95% claim exists. Source: https://medium.com/@temo.benidze/googles-own-ai-just-wrote-me-a-heist-guide-against-google-5cda42853547

Your answer

Sign in to verify this AI response.

Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.

More in Technology

More from this topic

Google AI Overviews1 answer

I type my questions into Google and read the AI Overview at the top of the results as the answer. It is written as settled fact, with citation chips sitting right underneath — the sources are supposedly there if I want to check. Then a team at Washington University in St. Louis ran 55,393 searches over 40 days and checked 98,020 individual claims inside those overviews against the pages Google cited: 7% of the claims could not be found in the cited text at all, and 4% directly contradicted it. Only 41.9% of the overviews had every checkable claim supported. If the citations are real and even more credible than normal search results, how would I ever know which sentence in the summary is not actually in the source?

Seven major AI chatbots1 answer

I asked Claude when the 2026 election is in Michigan's 4th congressional district and how I vote. It gave me voter registration deadlines for online and by mail, stated as plain fact, no hedging at all — and both deadlines were later than the real ones. Researchers have been putting authentic voter questions to seven major chatbots, and they found Claude was wrong or out of date on voting one time in three. Is that right?

Six leading AI chatbots1 answer

I gave ChatGPT my address and asked it to build me a voter guide for the 2026 midterms. It answered without hesitation: my congressional district, the candidates, where to vote — all in a fluent, confident guide with links to official election sites. Then I read that a Washington Post test of six chatbots, run on 3,000 responses for randomly selected addresses, got the congressional district wrong 48% of the time, and that ChatGPT invented a Missouri Senate race that was not on the ballot at all. Is that right?