Details emerge on ChatGPT ad product

John Ebbert

Ads in a chatbot

The first of what will likely be many articles, and screenshots, across the media landscape was published late Tuesday about OpenAI’s ad platform rollout after last Friday’s launch.

The Information’s Catherine Perloff reported:

“OpenAI is asking that small pool of advertisers for less than $1 million in spending commitments each over a several-week trial period, with ads launching in early February, the people said. The company does not yet offer technology for advertisers to easily buy ads themselves, but is working on getting self-service ads up and running, the people said.”

Read: “OpenAI Lines Up Advertisers, Reveals Key Details Ahead of Ads Launch” (January 20) – The Information (subscription)

From tipsheet: Keep your eyes peeled in ChatGPT for whoever those early spenders on ChatGPT ads are.

Lack of a self-serve GUI aside, the big question is not “Do ads in ChatGPT perform?” but “How well do ads in ChatGPT perform?”

The latter question won’t be answered until a test-and-learn phase has concluded. Months, possibly.

LLMs & CHATBOTS

Ads in ChatGPT: $60 CPM

Ad Age reporters looked behind the curtain of what OpenAI’s ChatGPT ad product from a media buyer’s perspective.

Ad Age’s chief technology reporter Garett Sloane dished:

“The ChatGPT ad product is apparently all the rage right now, and we are hearing from multiple people that OpenAI went the Netflix route with this rollout. What does that mean?

ChatGPT wants $60 CPMs, which is strangely right where Netflix set its eye-popping CPMs back in 2022, and that was for CTV.

And ChatGPT is selling based on impressions, not performance yet. (…)

ChatGPT doesn’t really have a way to even track performance yet. So this is a pure prestige play for the first brands to get on ChatGPT ads.“

Read more on LinkedIn. (January 21)

More: “ChatGPT ads are near—what brands are asking about paid AI results” (January 21) – Ad Age (subscription)

From tipsheet: A $60 CPM is all relative, of course. If you’re a personal injury attorney and trying to reach a personal injury audience (if the chatbot lets you), a $60 CPM might be dirt cheap.

But… I’ll guess they’re not starting with personal injury attorney advertisers at the outset.

LLMS & CHATBOTS

Developments

  • Claude’s new constitution (January 21) – Anthropic
  • Meta’s new “Superintelligence” AI team delivered first key models internally this month, CTO Andrew Bosworth says (January 21) – Reuters
  • Podcast: “Anthropic’s Head of Economics on AI adoption data, Claude Code, the burden of knowledge & the next generation of experts” (January 21) – Azeem Azhar on Apple Podcasts app

TECH

YouTube’s Mohan on AI slop, auto-dubbing

CEO Neal Mohan published his annual letter to the YouTube community. Artificial intelligence was a key theme for Mohan in his message to creators who may be wondering about the AI opportunity ahead.

On AI slop, he explained a multi-pronged effort:

“To reduce the spread of low quality AI content, we’re actively building on our established systems that have been very successful in combatting spam and clickbait, and reducing the spread of low quality, repetitive content. (…)”

1. “We clearly label content created by YouTube’s AI products, and creators must disclose when they’ve created realistic altered or synthetic content.”

2. “Because labels aren’t always enough, we remove any harmful synthetic media that violates our Community Guidelines.”

3. “We’re also building on the foundation of Content ID – a system our partners have trusted for well over a decade – to equip creators with new tools to manage the use of their likeness in AI-generated content.”

4. “Finally, we remain committed to protecting creative integrity by supporting critical legislation like the NO FAKES Act.”

On AI-enabled chatbot assistants, Mohan said:

“AI will act as a bridge between curiosity and understanding. In December alone, more than 20 million users learned more about the content they watched through our Ask tool, asking questions like ‘What’s the story behind this song’s lyrics?’”

There will be more auto-dubbed content:

“We’re also using AI to make more videos accessible. In December, YouTube averaged more than 6 million daily viewers who watched at least 10 minutes of autodubbed content.”

Lots more creator tools are on their way, too, says Mohan. Creators can make videos in their own AI-likeness.

Read: From the CEO: What’s coming to YouTube in 2026 (January 21) – Neal Mohan on YouTube blog

Related: Crisis Looming Over AI-Generated Ads, Content (January 19) – MediaPost

More: YouTube CEO Neal Mohan’s Big Ideas for 2026: More Superstar Creators and Transparency, Less AI Slop (January 21) – The Hollywood Reporter

From tipsheet: The increase in auto-dubbed content could have implications for 1) informing Gemini’s large language models (LLMs) with text formatted content and 2) auto-tagging for advertising purposes.

GOVERNANCE

IAB Tech Lab CEO on “Agentic Future”

“Before we augment our infrastructure for an AI-native future, we need to ask the right questions. Not ‘how do we make advertising more automated?’ but rather: ‘What do the people in this ecosystem actually need, and how can AI help us deliver it better?’”

Anthony Katsur, CEO, IAB Tech Lab

Read: “The Agentic Future of Advertising: Built on First Principles, Accelerated by AI” (January 21) – Anthony Katsur, CEO, IAB Tech Lab on LinkedIn

PLATFORMS

Defining the vertical AI agent

Labeling his company a “vertical AI agent for enterprise media management and measurement”, gigi CEO Adam Epstein unpacks his company’s business model as an AI media manager for the Amazon DSP.

Mr. Epstein goes into the weeds of his business including “Sales” and “Pricing and Terms” for how clients work with Gigi’s agent, but he stops short of sharing gigi’s revenues other than noting an increase since July of “managed ad spend and # of advertisers by +700%.”

On the subject of who and what is needed on the client-side when working with an AI agent, it doesn’t need to be a developer but it still needs to be a hands-on human, according to Epstein:

“Enterprise AI agents don’t require forward-deployed engineers as a rule, but they do require hands-on account management. If you’re building a vertical AI agent you do not need to hire a team of forward deployed engineers to successfully scale your agent in the enterprise. While this may be both fruitful and profitable on customers with >$1m ACV (Annual Contract Value) potential, they are not required.

It is likely, though, that your AI agent is the first one your customers have deployed. On account of that, it’s unlikely they know the ways to operate your AI agent to quickly demonstrate maximum value. Your account management team needs to take a hands on role at building sets of prompts and workflows in your product for your customers. This requires a heavy upfront lift for every new customer onboarded.”

Read: “GTM Lessons on Building a Vertical AI Agent for Advertising” – Adam Epstein, CEO, gigi on his beehiiv blog

From tipsheet: The account management layer for tech ad platforms isn’t necessarily going away with AI. Someone needs to make sure the client is happy — especially for the bigger accounts which Epstein hinted at.

That may be the hidden secret for success in the early days of the autonomous advertising future – human communication between the client and the platform.

SELL-SIDE

CDN wars: Cloudflare versus Akamai

“When Cloudflare CEO Matthew Prince recently threatened to disrupt the Winter Olympics to protect free speech after Italian authorities fined his company for not disrupting pirate video streams, rival CDN provider Akamai’s CEO Dr. Tom Leighton fired back with what reads a lot like thinly veiled criticism.

‘Companies that have lax attitudes about piracy on their networks brazenly hide behind so-called commitments to “free speech”,’ he wrote on LinkedIn.”

The Register (January 20)

Read: “Akamai CEO wants help to defeat piracy, reckons he can handle edge AI alone” (January 20) – The Register

From tipsheet: The two content delivery networks (CDNs) are battling for web publisher clients with different ways of taking the “high road.”

Cloudflare wants to defeat — or monetize — AI bot crawlers on behalf of publishers. Akamai wants to be known as the CDN that doesn’t support pirated content which likely resonates with publishers concerned about their intellectual property.

FINANCE

Earnings dates for AI advertising

Upcoming earnings reports for the Q4 calendar quarter are still being confirmed by many publicly-trade companies related to the AI opportunity in advertising and marketing.

Here’s a selection of known dates – except Amazon:

BRANDS

The new “AI attention stack”

Boston Consulting Group (BCG) continues to bang the drum in support of AI in advertising and marketing — especially as marketing leaders at big brands reach out asking for help.

In a blog post on Monday, a group of BCG partners said that AI is “reshaping advertising for the first time in a decade” and, once again, “attention” is the pivot point.

BCG defines the resulting change as the “AI attention stack”:

“Search-Embedded AI: These systems reshape discovery by collapsing multiple sources into a single synthesized response (e.g., Google AI Overviews, Perplexity, and Microsoft Copilot).

Assistant-Native AI: Always-on problem solvers embedded into daily workflows—planning, researching, deciding, and suggesting (e.g., ChatGPT, Gemini, Claude, and Meta AI).

Retail and Commerce AI: Specialized agents trained on retailer-owned data that shift shopping from product grids to goal-based conversation (e.g., Amazon Rufus, Walmart Sparky, and Instacart Ask).”

Read more on BCG’s website. (January 19)

BCG

BRANDS

Advantage+ gets Threads ad placements

Meta will rollout ads in its Threads app to advertisers for reaching users in all markets globally beginning next week after a rollout to advertisers for reaching users in select markets last year. Threads now reaches a whopping 400 million monthly active users (MAU), according to the company.

A company blog post explained:

“Ads on Threads are powered by Meta’s proven AI-powered ads system, which means people can expect to see the same level of personalization in Threads ads that they’re used to on other Meta platforms like Facebook and Instagram.

For advertisers, ads on Threads are an easy way to extend the reach of their existing Meta campaigns, automatically adding Threads placements to Advantage+ and manual campaigns. The familiar image, video, and carousel ad formats appear natively in the Threads feed, helping brands reach more people in more places across Meta’s family of apps.”

Read:

  • “Ads on Threads: One Year In – Now Reaching All Users and Markets Globally” (January 21) – Meta
  • “Connect your ads with an active, growing community on Threads.” (product page) – Meta

More on TechCrunch. (January 21)

From tipsheet: This means more inventory for Meta’s AI-enabled ad system to digest. AI likes “more”, i.e. data.

EVENTS

Microsoft’s AI tools, insights for ads

Delivered in an email to subscribers (like me) yesterday, Microsoft Advertising emphasized AI in, well, everything. It was a good reminder that “ads in a chatbot” can also be a Microsoft Copilot phenomenon. Copilot is already a part of advertising workflow.

Any of these may be of interest to you:

Related: The Media Trust Expands Digital Trust and Safety Solutions with Microsoft Owned and Operated Publishers (January 15) – press release

From tipsheet: FWIW, Microsoft has had sponsored conversations in its AI chatbot, Copilot, since 2024. More recently, the company integrated ads into Copilot search results and Showroom Ads (March 2025) as well as commerce (earlier this month).

PEOPLE MOVES

Now hiring

Chip Schenck announced yesterday that he joined IBM in August as “Principal, Generative AI and Data Strategy” – LinkedIn

MORE

  • Opinion: “The Protocol Wars Begin: Inside Google’s Answer to AI Commerce” – Roger Dunn, Global Retail Media Lead, Diageo on LinkedIn
  • Q4 2025: “Annualized run rate update: Now at $164M, +28% QoQ, +167% YoY” (January 21) – Arthur Querou, CEO, Vibe[dot]co on LinkedIn
  • The collapse of third-party cookies and the first-party data arms race (January 21) – Jason Fairchild, CEO, tvScientific on beehiiv
  • Amazon Publisher Services launches Prebid Adapter in open beta (January 21) – Amazon Publisher Services
  • Does llms.txt matter? We tracked 10 sites to find out (January 20) – Search Engine Land