AI Change Tracker
4

404 Media finds ChatGPT ads can be targeted by phone number, email, and prompt content

2026-10-15

404 Media's own test campaigns show OpenAI lets advertisers target specific ChatGPT users by uploading raw or hashed email addresses, phone numbers, or Android advertising IDs, which OpenAI checks against real accounts (uploaded lists under 25,000 entries can only be used for exclusion, not inclusion). Ads can also be targeted using advertiser-supplied 'context hints' tied to the topics or prompts users type into ChatGPT. OpenAI did not confirm whether it hashes raw PII uploads the way Meta does. DecryptAds co-founder Zach Edwards characterized the rollout as OpenAI 'boiling the frog' on privacy. ChatGPT ads launched in the US in February 2026 and expanded to image-generation results this month.

Significance 4: 404 Media's own hands-on testing documents previously undisclosed ad-targeting mechanics (PII matching, prompt-based targeting) at OpenAI's ~1B-MAU scale, a verifiable primary finding rather than opinion, materially updating OpenAI's monetization/privacy posture.

business operational Lab strategy Policy / regulation

Implications · machine-drafted, not owner judgment

This hands OpenAI's roughly one billion ChatGPT users into a PII-matched ad system many don't know exists, arriving the same week OpenAI faces an eight-state AG lawsuit over product-design harms — a concrete, technical addition to the stated-vs-revealed record this tracker has been building on OpenAI specifically. For enterprises evaluating ChatGPT Enterprise/Teams privacy posture, this is now a specific detail to raise in due diligence rather than speculation, and it narrows the credibility of any claim that chat content is walled off from commercial use.

Watch for
  • Whether OpenAI confirms or denies hashing raw PII uploads the way Meta does
  • A regulatory inquiry opening specifically on ChatGPT ad-targeting practices
  • Confirmation of whether paid tiers are exempt from this targeting

Sources

← Back to feed