Monday Briefing: ChatGPT Ads Hit Europe This Morning

Good morning, it's James here. Six months after the U.S. pilot, ChatGPT starts serving ads across Europe today. That is the headline. The more interesting story is the fight over who gets to stand between an AI agent and your customer.

(Was this newsletter forwarded to you? Sign up here.)

The Lead: ChatGPT Ads Hit Europe This Morning

What happened. OpenAI begins serving ads in ChatGPT across 31 European markets today, including Germany, France, Spain, Italy, Poland, the Nordics, Ireland and the Netherlands (Digiday). Buying runs through Publicis, Omnicom, WPP, Havas, Dentsu and MediaPlus. Self-serve for small advertisers is "coming," with no date. The pilot had been limited to the U.S., Canada, Australia, New Zealand, the U.K., Japan, South Korea, Brazil and Mexico (Adweek).

OpenAI's own numbers. Ad revenue up more than 25% since the start of August, 900 million-plus weekly active users, roughly 20% of queries showing direct commercial intent, and cost-per-click bidding now accounting for most of the spend (Digiday). Colin Fleming, OpenAI's enterprise CMO, called it a "material expansion" (Adweek).

The Europe twist. OpenAI is asking users for consent before personalizing. Free and Go users choose between personalized ads and contextual ads that use only the current conversation and rough location. Paid tiers stay ad-free (Digiday). Meta spent years and a lot of fines fighting that same requirement. OpenAI is skipping the fight and building the consent wall on day one.

Why CMOs should care. CPC bidding dominating spend tells you the early buyers are performance teams treating this like search. Twenty percent commercial intent on 900 million weekly users is a bigger commercial surface than Bing ever was. And in Europe a meaningful share of your impressions will be contextual, not behavioral. Conversation-level relevance is the targeting. Your product data and your copy carry the load your audience segments used to.

The take. If you run Europe and you buy through one of those six holdcos, you can be live this week. Do it small, do it with a clean product feed, and set up a holdout before your agency sends the first excited screenshot. Nobody has a credible attribution story for this channel yet, including OpenAI. The brands that build their own will own the benchmark when the self-serve crowd arrives.

The Debate: Who Owns the Agent's Front Door

Two fights this week, same question: when an AI agent shows up on behalf of a customer, who sets the terms?

Round one: Perplexity versus Time. Three weeks ago we covered Time selling ads inside the markdown versions of its pages built for AI crawlers. Perplexity has now blocked those ads from influencing its index and warned that publishers doing the same "risk reductions to their reputation and trust scores" (Marketing Tech News). The objection is not the ads. It is serving one page to humans and a different one to machines, which Perplexity calls deceptive even when the placement is labeled.

Round two: Amazon versus Perplexity. Amazon has asked the 9th Circuit to reinstate its ban on Perplexity's Comet shopping agent. Its filing argues the lower ruling "degraded website owners' ability to set the terms on which powerful, fast-evolving, and potentially destructive AI agents may enter their secure systems," and that when a user says "order paper towels," it is Perplexity's reasoning layer, not the user, that walks into the Amazon store (MediaPost).

Why it matters. Perplexity is the enforcer in one case and the defendant in the other, and its position is consistent. Agents should see what humans see, and agents should be allowed in. Amazon's position is just as consistent. Our store, our rules. Every retailer and publisher with a first-party relationship is going to have to pick a side. If you sell through a marketplace, the marketplace is deciding for you right now.

Sponsored By:

Winning Startups Aren't Bigger. They're Leaner.

Founders deploying AI agents across sales, marketing, and customer service are closing more with smaller teams. 65% more sales leads. Less headcount.

Download the free Practical Guide to Agentic GTM for Startups and start building the stack your competitors dream of.

Get the Free Playbook

By The Numbers: Eighth of Eight

Eighth. That is where follower count landed when CreatorIQ asked brands to rank what matters in a creator partnership. Dead last. Brand suitability ranked first, ahead of content performance. And 53% of creators with at least 500,000 followers say they feel tension between what their audience wants and what brands demand. The survey covered 5,095 respondents across 100 regions, fielded May 29 to June 29 (Marketing Dive).

The second stat explains the first. The bigger the creator, the more your brief fights their audience. If your influencer scorecard still starts with reach, you are paying a premium for the creators most likely to resent the work.

The Reading List

• Mark Ritson: I used Claude to help write this. It's still mine. Right? (The Drum). "A fingerprint isn't authorship." The best argument yet for judging work by who made the decisions, not by which tool left traces.

• The best advertising advice Amazon gives isn't in a slide deck (The Drum). Kiri Masters reads three years of Sensor Tower data on how Amazon spends its own $2.69 billion. Go where the reach is, pay up for the format that works harder, keep a small bet on what's next.

• Creators are moving into brands' AI search playbooks (Digiday). Zoom hired a creator-journalist to build LLM authority. Viral Nation's Joe Gagliese explains why. The machine wants "the most accessible public information and the most authentic," and "that's not coming from the brand themselves."

• IAB revisits AI disclosure in ads as legal requirements multiply (Marketing Dive). Version two of the framework, now that disclosure laws are live in the EU, Asia, New York and California. IAB's Caroline Giegerich puts the tension in one sentence, "labeling everything teaches consumers to ignore labels."

• Google releases August 2026 spam update (Search Engine Land). Third of the year, global, all languages. If organic moved last week, check this before you blame the content team.

One More Thing

Six months from U.S. pilot to 31 new countries and most of the spend already on CPC. Google took years to get search ads to that point. The platform cycle has compressed, and your measurement cycle has not. That gap is where the next two years of marketing budget gets won or lost.

See you next Monday. Make it count.

—James