OpenAI to watermark ChatGPT text, insert visual ads
OpenAI responds to regulatory pressure in Europe as it simultaneously seeks new revenue through ads
Open AI is to introduce invisible “watermarks” into text generated by ChatGPT and Codex in order to comply with the EU AI Act. At the same time, it is expanding advertising in ChatGPT as it seeks new sources of revenue.
Taken together, the two announcements show OpenAI responding to increasing regulatory scrutiny in Europe as well as growing demands to monetise its services at scale and reduce its losses.
In a post on its website, OpenAI announced that it will add watermarks to ChatGPT and Codex text output in the EU over the coming weeks using its "textGrain" system. All API customers can now opt in, as of 5th October, although the feature is switched off by default. The company said it is not making text watermarking a global default at launch.
textGrain subtly influences word choices in the model’s output in a way that is imperceptible to humans, but which can be analysed by a detector to check whether it is the work of AI.
The move is intended to comply with the EU AI Act's requirement that AI-generated content be machine-identifiable.
Detection is not perfect, with quality dependent on the length of the text and other factors, while edits after the fact can disguise the use of AI. In tests on a 400-token passage, OpenAI says that replacing just 10% of words with synonyms reduced detection rates from about 92% to 66%.
Nevertheless, it represents one of the first examples of the EU AI Act forcing a product change on a leading AI vendor. It follows recent moves by the European Commission to place ChatGPT, Reddit and Roblox under the EU's strictest digital safety rules.
OpenAI’s main rival, Anthropic, began adding watermarks to text and images generated by Claude models back in August, after signing the EU AI Act's Article 50(2) Code of Practice on transparency of AI-generated content.
Visual ads come to ChatGPT
In a move to boost its income ahead of a planned IPO next year, OpenAI is introducing a new visual advertising format for ChatGPT. Initial tests will be run in the US later this month, with ads appearing during image-generation sessions.
“Ads will be clearly labeled, and remain separate from the image being created,” OpenAI promises, in an explanatory post.
In a sign that there will be more of this to come, OpenAI is also expanding its advertising measurement, attribution and brand-safety partnerships.
It comes at a time when enterprises are increasingly concerned about rising AI costs, while AI providers raise prices in an effort to recoup huge investments in infrastructure.
In 2024, OpenAI CEO Sam Altman declared "I hate ads," describing advertising as a "last resort" for the company. "Ads plus AI is sort of uniquely unsettling to me,” he added.
In January 2026, the company announced it would be showing ads to some users in the US on ChatGPT's free tier and Go subscriptions.