Summary: Senator Edward Markey wrote on January 22, 2026 to seven AI companies, including OpenAI, Anthropic, Google, Meta, Microsoft, Snap, and xAI, asking how chatbot ads would work and what safeguards exist, with replies due February 12. The probe followed OpenAI’s January 16 announcement of ad testing in ChatGPT’s free and Go tiers. He raised privacy, young-user safety, and manipulation risks from using sensitive conversation data for targeting.
January 22, 2026 is the day Senator Edward Markey (D-MA) wrote separately to the chief executives of OpenAI, Anthropic, Google, Meta, Microsoft, Snap, and xAI, telling each to lay out how it plans to run ads in its chatbots, plus what safeguards for users it already has; replies are required by February 12. What set it off was OpenAI itself: on January 16 it had announced ad testing for the free and Go tiers of ChatGPT.
Why the letter counts: timing and placement. Hundreds of millions of people now lean on chatbots when they need information or advice for decisions, and whichever shape ads take inside the interface will settle whether what emerges is a fresh business model — or the opening of another covert-manipulation cycle. Stepping in ahead of the launch, Markey cast it as a consumer protection problem at industry scale.
The Backdrop: ChatGPT Ad Testing Starts January 16
OpenAI broke the news on January 16: ad tests would start inside ChatGPT for free-tier users and the lower-cost Go plan. The Verge’s report describes the format: the ads surface as “sponsored” products and services toward the foot of a conversation, matched to what the chat is about; anyone under 18 is left out, and so are conversations touching physical health, mental health, or politics. The rationale came from spokesperson Liz Bourgeois, who put the ads down as “about supporting the resources needed to keep this technology freely accessible,” with OpenAI to be “guided by transparent principles designed to protect trust rather than optimize for engagement.”
The Questions Markey Poses: Privacy, Young Users, Manipulation Risk
Every company must account for how its ad placement works, whether what users say in a chat can end up steering ad targeting, plus what specific mechanisms exist to keep users protected. Chatbot ads embedded in the conversation, Markey wrote, “raise significant concerns for consumer protection, privacy, and the safety of young users,” and he pressed on the danger that companies could take people’s “personal thoughts, health questions, family issues, and other sensitive information” and build ad targeting on top.
The argument then moves onto relational ground: people grow emotionally attached to these systems, and companies could use that attachment to “prey on the very relationships their systems have fostered.” He also cautioned against letting chatbots “become another digital ecosystem structured to covertly manipulate users.” One line from the press release holds the entire position: consumers deserve to know whether the chatbots they use are truly helping them — or quietly feeding them blurred advertising.
Independent Answers: Simple to Pledge, Hard to Prove
Three commitments make up OpenAI’s public position on ads: clear labeling, privacy protection, answer independence. The questions from Markey aim straight at how the promises get carried out — The Verge highlights that he homed in on whether material gathered in conversations on sensitive topics could end up shaping the ads shown in later ones, whatever limits the company says it already applies.
The truly hard part sits in the engineering. A line drawn at the level of system design can keep “ads don’t affect answers” true, but “ads relevant to the conversation” cannot work unless the conversation gets read first. Where relevance stops and the manipulation begins is settled only by mechanisms that can be audited, plus verification from outside — not a page of principles.
Seven Letters, One Industry Message
What deserves attention is the breadth of the recipient list: OpenAI is not the only addressee — Anthropic, Google, Meta, Microsoft, Snap, and xAI all got letters as well. In Markey’s reading, this is the business model a whole sector is converging on — not the experiment of a single company. When February 12 comes and goes, the seven replies on “does conversation data feed ad targeting” and “what protects minors” become the groundwork for whatever Congress does next and for the legislative drafts that follow.
Teams building products should read the letter as advance notice: assume from day one that whatever money-making mechanics go into a chat-based interface will one day have to be laid out for regulators and Congress — the data flows, the targeting logic, the user protections. Advertising is, at present, among the AI industry’s favorite ways to commercialize (if you want the wider picture, we charted it in our 2026 opening outlook: /blog/ai-2026-opening-outlook/), yet no product choice currently carries more regulatory risk.