ChatGPT Ads Have No Minimum Spend Now. Here's Whether That's Actually Good News for You

Here's what actually changed, and who it's for right now. ## What Actually Changed On May 5, 2026, OpenAI quietly removed the biggest barrier to testing ChatGPT ads: the $50,000 minimum spend. Since then, self-serve access has been open to anyone in the US, including small businesses and startups, not just the big agency holding companies who got early invite-only access under a $200,000 pilot commitment (Digiday). Eligible categories right now are household and consumer goods, local services, travel, entertainment, digital products, and education. Outside those categories, you're not eligible yet; OpenAI says it will expand as its review and compliance systems mature. What you actually get today is modest: a self-serve Ads Manager still in beta, cost-per-click bidding only, a conversion-tracking pixel, and a Conversions API still in development. The ad format, a favicon plus a line of text, isn't winning any creative awards. CPA bidding and third-party measurement have both been promised publicly, but OpenAI's ads lead has declined to give a timeline or name measurement partners (Digiday). So for now you're bidding blind on attribution, using your own UTMs and conversion tracking as the source of truth, not theirs.  ## The Number That Should Give You Pause Here's the number that should make any operator pause before committing real budget. A Quad/Harris Poll survey of 2,180 US adults, run in February 2026, found 75% would trust an AI shopping agent less, and trust the brand less, once they learned a recommendation was paid for (Quad). ChatGPT's entire value to users right now is that it feels like an advisor, not a search results page cluttered with sponsored links. The moment that erodes, the channel's premium, people actually acting on what it tells them, erodes with it. Early advertisers already in, per public reporting, include Target, Williams-Sonoma, Albertsons, and The Knot Worldwide, alongside agency-placed brands like Adobe, Ford, and Mazda (Floodlight), so bigger players are already staking a claim on the "safe" categories.  If you're in an eligible category, here's what to actually do: set aside a small test budget, the kind you'd spend to learn, not to scale, run CPC only, build your own landing-page tracking since you can't rely on their attribution yet, and check back in on CPA bidding and measurement partners before committing anything larger. If you're outside those categories, spend this quarter watching, not testing, and put the saved effort into channels where measurement already works. Same principle either way: don't confuse "underpriced attention" with "safe to scale" until the platform's own numbers catch up to what it's asking you to trust it with.