OAI-AdsBot
OAI-AdsBot is OpenAI's bot for checking the pages advertisers submit as ads on ChatGPT. It only visits those pages.
OAI-AdsBot is OpenAI's bot for checking ad pages. When an advertiser submits a page as an ad on ChatGPT, OAI-AdsBot visits it to check that it's safe. According to OpenAI's bots page, it only visits pages submitted as ads, and what it collects isn't used to train OpenAI's models.
If you don't advertise on ChatGPT, you're unlikely to see it at all. If you do, it's the bot that has to be able to read your landing pages.
How it differs from OpenAI's other bots
OpenAI gives each of its bots a job and a name, so you can treat each one differently:
- OAI-SearchBot reads pages so ChatGPT can cite them in search answers.
- GPTBot reads pages that may be used to train OpenAI's models.
- ChatGPT-User visits a page when someone using ChatGPT asks it to.
- OAI-AdsBot checks pages submitted as ads.
That separation matters when you write your robots.txt or firewall rules. Blocking GPTBot keeps your pages out of training, and it doesn't touch OAI-AdsBot. Blocking every bot with "OAI" or "GPT" in its name can catch more than you meant to.
An example
Say Harrow & Pine starts advertising on ChatGPT and points its ads at a new sofa range. Its security team has a firewall rule that turns away unknown bots, and OAI-AdsBot is on the wrong side of it. The landing page can't be checked, and the ad has nothing to send people to. Allowing the bot through, for the ad pages at least, is the fix.
Watch for fakes, too. Anyone can give their bot a known bot's name, so check a visit's address before you trust the name.
In AnswerOrbit
Ads shows which brands advertise next to ChatGPT's answers to your prompts, how often, and who books their ads. The Audit tab in Website lists each AI bot we know with its purpose and whether your robots.txt allows it. When you connect your server logs, we only count a bot's visit after checking its address, and requests that fail the check show as Unverified requests.