OpenAI plans visual ads during ChatGPT image generation

OpenAI plans to test a new visual advertising format during ChatGPT image generation in the United States later this month, starting with an initial group of advertisers. Its October 5 announcement does not specify which user plans will be eligible. S1

For someone using AI to explore an idea, the important question is how clearly the product separates help from selling. A useful test would establish whether people understand what they are seeing, why it appeared and how to change it. Those are user-experience questions as much as advertising questions.

What users would see

OpenAI says the ads will be labeled and kept separate from the image being created, and that advertising will not influence ChatGPT’s answers. These are company commitments; the announcement is not a completed evaluation. S1

Consider an illustrative task: sketching ideas for a living room. A commercial suggestion appearing during that process could be useful, distracting or easy to mistake for part of the assistant’s advice. The label, placement and timing would matter. This is an example, not a reported test result.

What the existing controls mean

For its broader advertising system, OpenAI’s help documentation describes controls to dismiss or report an ad, inspect why it appeared, switch off personalization and clear ads data. It says disabling personalization still allows selection based on the current conversation, while past chats and memory can contribute when relevant settings are enabled. S2

The same documentation says advertisers do not receive private chats or memories. That describes a boundary on advertiser access; it does not mean conversation context plays no role in selecting an ad. S2

Readers should keep those questions separate: who receives information, what the platform uses internally and which choices are available. Bright’s conversation-privacy explainer takes the same practical approach to understanding where a conversation goes and who can access it. S5

What the next evidence should show

OpenAI’s placement policy excludes sensitive contexts, including personal-health and emotionally vulnerable conversations. It also requires ads to be distinguishable from ChatGPT’s own interface. These are published standards, rather than proof of consistent enforcement. S3

The company is developing brand-suitability pilots with DoubleVerify and Integral Ad Science. OpenAI describes a controlled testing environment in which those partners would assess safeguards without accessing private user conversations. The announcement supplies no completed pilot findings. S4

Published results could help, particularly if they explain the scenarios tested and the failures found. A brand-suitability assessment would still need to be read for its actual scope before drawing conclusions about the whole user experience.

OpenAI also announced measurement integrations that let advertisers send conversion data from their existing systems, alongside work on experiments intended to assess advertising’s causal impact. Those tools concern business outcomes; by themselves, they cannot establish that users find the new format helpful or clear. S4

As testing begins, the useful checks are straightforward: Can people reliably identify the paid placement? Can they find and understand the controls? Are sensitive conversations excluded in practice? Does the ad leave room for the task they came to complete?

Bright’s view: commercial discovery may have a place in an AI assistant, but the case for it should include evidence about people’s understanding and control. The forthcoming test is a chance to start providing that evidence.