Episode 13 - The Thinking Behind Ads in ChatGPT

How should advertising work in an AI product? Asad Awan, one of the ad leads at OpenAI, walks through how the company is approaching this decision and why it’s testing ads in ChatGPT at all. He explains how ads are built to stay separate from the model response, keep conversations with ChatGPT private from advertisers, and give people control over their experience. Chapters 00:00:29 — Mission and principles 00:04:01 — Separation between ads and answers 00:07:31 — Who will see ads 00:08:52 — Internal input and decision-making process 00:11:06 — Controls and how ads will work 00:15:53 — Guardrails for sensitive conversations 00:17:33 — Skepticism about ads 00:20:26 — Helping small businesses 00:24:13 — Future of ads

Asad Awanguest

Feb 8, 202625mWatch on YouTube ↗

WHAT IT’S REALLY ABOUT

OpenAI outlines trustworthy ads in ChatGPT without influencing answers ever

Asad Awan explains why OpenAI is adding ads to ChatGPT: to fund higher free-tier usage limits and broaden access in line with the mission to bring powerful AI to everyone.
He emphasizes a hard separation between ads and model answers—both visually and technically—so the model does not know what ad is shown unless the user explicitly chooses to ask about it.
The ad system is framed around a rubric prioritizing user trust over user value, advertiser value, and revenue, with strong privacy promises (no sharing conversations with advertisers; no ads in sensitive conversations).
Ads will initially be limited and shown only to Free and Go tiers, with Plus/Pro/Enterprise remaining ad-free, and with extensive user controls for personalization and data use.

IDEAS WORTH REMEMBERING

5 ideas

Ads are positioned as a funding model for broad access.

OpenAI argues ads help deliver “the best version” of ChatGPT to hundreds of millions of users by supporting higher limits and more capable experiences for people who can’t or won’t pay.

Answers and ads are designed to be independent by default.

The model won’t be trained/conditioned on the ad being shown and won’t even know an ad exists in the UI. Users can optionally click “Ask ChatGPT about this ad” to bring it into the conversation context.

Trust is explicitly ranked above revenue in decision-making.

Awan describes a simple-but-strict rubric: user trust > user value > advertiser value > revenue. The example given is avoiding “creepy” relevance that makes users suspect surveillance even if the ad performs well.

Ads will not appear for paying tiers and enterprise contexts.

Ads are planned only for Free and Go users; Plus, Pro, and Enterprise remain ad-free, reflecting different product contexts and expectations—especially around enterprise data stewardship.

Users get meaningful controls over ad personalization and data sources.

Controls include viewing what data is used for ads, choosing whether past chats can be used (with sensitive chats never used), clearing history, turning off personalization, and opting out of ads entirely by upgrading.

WORDS WORTH SAVING

5 quotes

We are in the business of trust.

— Asad Awan

The answers need to be independent from the ads—both visually, but also in how the models are trained and how the system works.

— Asad Awan

Your conversations are private… the conversations are never shared with advertisers.

— Asad Awan

Is creepy okay if it is good? It’s not.

— Asad Awan

User trust is the most important thing. User trust more than user value… more important than advertiser value… more important than revenue.

— Asad Awan

Mission-driven rationale for adsWall between ads and answers (visual + technical)Tiering: Free/Go vs Plus/Pro/EnterpriseInternal governance and decision rubricPersonalization, transparency, and user controlsSensitive-topic detection and ad exclusionFuture: conversational and agentic discovery/marketplaces

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.