ChatGPT ads for Europe

John Ebbert

ChatGPT ads in Europe

The contextual nature of large language models (LLMs) may enable a new conversational ad format. But on privacy, the old rules are reasserting themselves.

Digiday’s Krystal Scanlon and Seb Joseph reported on Friday: “OpenAI appears to be getting ready to expand its ads business to Europe. A code update to its conversion tracking pixel, which Digiday reviewed, points to the company building the technical groundwork needed to run advertising in the European Union. The update adds a consent management system, essentially a mechanism that lets advertisers ask users for permission before tracking them, and stops tracking if that permission is withdrawn.”

Read: “OpenAI starts laying foundations for ChatGPT ads in EU” (May 1) – Digiday (subscription)

From tipsheet: Conversational advertising runs into General Data Protection Regulation (GDPR) reality. Even if ChatGPT ads lean contextual, the moment a pixel and conversion tracking enter the picture, consent becomes table stakes. LLMs can target without tracking, but they can’t optimize without feedback.

LLMs & CHATBOTS

Developments

  • Meta buys robotics startup to bolster its humanoid AI ambitions (May 1) – TechCrunch
  • AI outperforms doctors in Harvard trial of emergency triage diagnoses (April 30) – The Guardian
  • xAI launches Grok 4.3 at an aggressively low price, along with a fast, powerful voice cloning suite (May 1) – VentureBeat

PLATFORMS

Ads and Amazon’s terms for third-party agents

On last week’s earnings call, CEO Andy Jassy made a notable comment on agentic commerce that goes directly to Amazon’s ad model.

Jassy said on Wednesday (lightly edited):

“… we’ll do a lot of work with third-party horizontal agents to try and make that customer experience better. It reminds me of the early days of search engines, and they’re trying to refer business to e-commerce — you know, it has never been a giant part of the referrals to our e-commerce business. But over the years, the experience got better.

What you see with agentic commerce is a small fraction of what we see with the search engine referrals. The experience just hasn’t gotten great with these third-party horizontal agents yet. They’re not often able to get the pricing or product information right. They don’t have any personalization data or shopping history. We do want to see that get better with third-party horizontal agents. We’re having conversations with all those folks to try and make that better and find something that works for customers and all the companies. It’ll be interesting over time which agents customers choose to use.”

For Amazon, this isn’t just about traffic. It’s about whether that traffic still flows through the surfaces where ads are served.

Discovery shifts upstream

And to be clear, referrals from third-party agents mean referrals from systems spanning AI search, chat interfaces, and assistants like Google’s AI search, ChatGPT conversations and Siri.

As Stratechery analyst Ben Thompson noted, “I was surprised to see Jassy say that the company wanted to figure out how to work with third-party agents.” Thompson’s point is that Amazon has historically benefited from owning the starting point of shopping, not outsourcing it — which is also where Amazon monetizes that intent through ads.

And here is the first of two critical distinctions: Amazon’s traditional third-parties are participants inside the marketplace (sellers, advertisers, developers). Jassy’s “third-party horizontal agents” are systems outside the marketplace that refer users in (like “early search”).

Perplexity

The second distinction has to do with Amazon’s current battle with Perplexity over Perplexity bots transacting on Amazon properties.

Perplexity is not a horizontal agent in the way Jassy is describing. A true horizontal agent sends users to Amazon, acts as a referrer and leaves the discovery and the transaction to Amazon.

Perplexity’s model wants to go further by deciding what to buy as well as executing the purchase, which turns Amazon into a transaction endpoint — removing the shopper journey (browsing) layer where Amazon collects data, optimizes performance and serves ads.

That’s the difference between an agent that refers demand and one that captures it.

So, this makes Perplexity’s solution a competing interface, not a complement as Jassy envisions with systems like ChatGPT.

This dynamic helps explain Amazon’s support of Universal Commerce Protocol (UCP) last week. Amazon wants in to agentic commerce on its terms with no disintermediation of the consumer.

Read:

  • Amazon.com Q1 FY2026 earnings call transcript (April 29) – Yahoo Finance
  • Amazon Earnings, Trainium and Commodity Markets, Additional Amazon Notes (April 30) – Ben Thompson on Stratechery (subscription)
  • Last week, Digital Content Next filed an amicus brief in support of Amazon and its lawsuit against Perplexity (May 1) – Jason Kint, CEO, DCN on X

From tipsheet: This is ultimately a fight over where the decision to buy happens and where ads can play a part on Amazon properties.

Amazon’s model depends on users starting on Amazon and browsing results which includes seeing sponsored products. Jassy’s “horizontal agents” idea preserves that model by acting like upstream demand sources.

If agents become the interface, ads don’t disappear. They move from placements in a marketplace to recommendations inside a decision. (e.g., Amazon’s Rufus chatbot)

That helps explain why publisher groups like Digital Content Next, led by Jason Kint, have taken a similar stance toward AI intermediaries. It’s not that publishers are “pro-Amazon.” It’s that both are pushing back on agents that sit in front of the interface, capture intent and extract value.

For publishers, that threatens traffic and ad revenue. For Amazon, it threatens product discovery and retail media.

In both cases, the questions are the same: 1) who controls the decision of what to buy and 2) who gets paid when that decision is made.

MARKETING

AI, marketing and proving it to the C-suite

“One of my goals at Possible was simple: find real marketing tactics from people actually doing the work. So I tracked down Carolyn Everson, Vinny Rinaldi and Julia Fedor — leaders driving transformation at Coca-Cola, Hershey and United.

Here’s what stood out:

  • AI ‘pilots’ are everywhere in marketing orgs right now. But at United, a storytelling edge comes from an actual pilot—Paul Holte—who’s building a following on social. Holte and Fedor spoke at Portrait Media Group’s space, and gave a lesson in how brands’ social media is changing, finding ways to engage by combing the daily feed, including a fun story from Reddit.
  • Carolyn Everson, on stage with Coke’s Manolo Arroyo, made the C-suite case plainly: measurement is evolving. Coca-Cola is now thinking in terms of weekly active consumers—a shift toward always-on engagement. Coca-Cola is using its data advantage to glean insights from the kiosk to the checkout line, to learn about its audience…”- Garett Sloane, Senior Editor of Technology, Ad Age on LinkedIn (May 1)

Read more on LinkedIn. (May 1)

More: How Coca-Cola, Hershey and United are using AI and data—and proving the value of marketing to the C-suite (May 1) – Ad Age (subscription)

TECH

Meta Business executives on the move

Meta is seeing movement across key roles tied to its business AI and messaging strategy.

Clara Shih, who was Meta VP and helped launch new Meta Business AI products during Advertising Week in October (read tipsheet), recently left the company.

And Nikila Srinivasan, Meta’s VP of product management for business messaging, has also left the company but will remain as an advisor through the summer. You may recall that she was the Meta point person for the WhatsApp launch of business messaging at Cannes last June.

From tipsheet: These departures also land amid cost-cutting at Meta tied to its AI push. CEO Mark Zuckerberg recently confirmed plans to cut roughly 10% of the workforce (about 8,000 roles) as the company ramps spending on AI infrastructure, with the possibility of further reductions still on the table.

Taken together, the signal isn’t that Meta is pulling back from AI. It’s the opposite.

As Meta reiterated on last week’s earnings call, it’s investing heavily in AI infrastructure and models while still figuring out how to translate that into business-facing products like messaging, automation and ad tools.

Leadership turnover in Business AI and messaging suggests the advertiser-facing layer of Meta’s AI strategy is still being worked out.

Also in the mix: Manus raising additional questions

A few open questions shaping Meta’s Business AI roadmap:

  • Was the idea with Manus when it was acquired in December to take over development of all things B2B workflow? Hence, layoffs in Meta Business?
  • Now that China is reportedly taking back the startup and its founders, are new plans required for Meta and Business AI?
  • Or is the Singapore version of Manus (unreachable by China?) able to fulfill Meta’s original Manus plans?
  • Finally, where does last week’s MCP announcement by Meta fit in all this? The company says it wants to enable third-party AI business tools.

FINANCIALS

Earnings reports this week

TECH

Bill Wise ‘amps’ Mediaocean’s AI positioning

Bill Wise, CEO, Mediaocean, wrote on LinkedIn on Friday:

  • “The last few months have brought a wave of questions about the so-called SaaSpocalypse and I understand why. AI is putting real pressure on the application layer of software. It’s becoming easier to generate interfaces, automate workflows, and replicate features that once looked durable. In advertising, as in many industries, that raises an obvious question: if software becomes easier to reproduce, what remains defensible? And, more importantly, how does that translate to better outcomes?”
  • “My view is that, when it comes to ad tech, the real moat was never just the software. It’s the infrastructure beneath it. Thomas Edison did not win because he invented the best light bulb. Others were working on that too. What mattered was the system that made electric light usable at scale: generation, wiring, distribution, and control. In other words, the grid…”

Read: Powering the grid to light up AI in advertising (May 1) – Bill Wise CEO, Mediaocean on LinkedIn

PROTOCOLS

Podcast: Index Cloud’s Performance Max moment

On the Marketecture podcast, Marketecture’s Ari Paparo brings his ad tech product expertise to a conversation with Index Exchange CEO Andrew Casale about the newly-announced Index Cloud.

Central to Index Cloud is “containerization” which underpins IAB Tech Lab’s protocol, Agentic Real-Time Framework (ARTF).

Casale discussed his vision for Index Cloud and the coming efficiency unlock (lightly edited):

Ari Paparo: “If you were to prognosticate three to five years out, assuming programmatic still exists, what do you see this cloud looking like?”

Andrew Casale: I think programmatic won’t just exist. I think it will be bigger and better in three to five years than it is today. And I think it’s innovations like this that will make that possible.”

“I’ll give you a few versions of the future that I see. The one obvious one being, I think very few bid requests will ever leave the exchange in five years. There certainly might be obvious cases, big giants that will have their own infrastructure.
But I think the vast majority of bidders will containerize because there’s a huge business benefit. (…)”

“The second one is what kicked off the portability of models or the custom bidding optimization party was about five milliseconds in the hot path. Imagine a world in the not-too-distant future in a few years where, instead of giving a custom bidding algorithm five milliseconds, if the bidder is so fast, we can give the algorithm a lot more time.(…)”

“We then get to a place where I think we can have a whole lot of companies building things that start to feel a lot more like Google’s Performance Max.”

“And it can get really fun in our ability to try new techniques at massive scale with a huge Internet firehose with low costs to start to find novel ways to drive outcomes at scale. That’s the hope that I have for the market, and that’s what I think this is all about — perform better and drive better outcomes.”

Listen: “Episode 171: Dispatch from Possible plus Andrew Casale on Index’s cloud computing initiative” (May 1) – Marketecture Podcast

From tipsheet: Great, and important, discussion between two ad tech OGs.

Casale is describing a future where the open web gets its own version of Google’s Performance Max logic: more decisioning, optimization and outcome-seeking models running closer to the exchange.

That is why containerization (and Amazon’s recently-donated DTE) matters. If bid requests no longer need to travel across the programmatic supply chain, the exchange becomes more than a pipe. It becomes a computing layer where buyers can run custom models against the open web’s firehose with lower latency and potentially lower cost.

That vision aligns with what Chalice AI CEO Adam Heimlich has consistently argued while building Chalice AI: the next phase of programmatic may be less about adding intermediaries and more about where the model actually runs.

Related: “The Container Explainer” explaining ARTF (February 2) – Adam Heimlich and Gareth Glaser on the AdTech AdTalk podcast on YouTube

PROTOCOLS

Bedrock engineer parsing the protocols

Damian Naglak, Head of Engineering for DSP Bedrock Platform, is continuing to attract attention on LinkedIn for his astute observations on, and work with, advertising protocols including Ad Context Protocol (AdCP), ARTF and the protocol compendium of AAMP.

(Bedrock is also the first bidder instance within Index Cloud.)

Snippets from recent posts by Mr. Naglak include:

  • “An agent moves a deal from quoted to booked to delivering. Two weeks later, finance flags a discrepancy on the rate. The defensible answer can’t be ‘the agent decided’. It has to be a list of transitions: timestamps, actors, reasons. And that’s the gap AAMP 2.0’s buyer and seller SDKs are built around.” (May 1) on LinkedIn
  • “Every time an exchange wanted to add a new vendor capability, it used to mean a custom integration. Fraud detection, audience activation, dynamic floors, bid shading: one API per vendor, one payload per vendor, one rollout timeline per vendor. IAB Tech Lab‘s ARTF v1.0 replaces that…” (April) on LinkedIn
  • “A typical creative workflow in AgenticAdvertising[cot]org has four actors: a buyer, a creative agent, multiple sellers, and a brand.json at a known URL. No single actor owns the full flow…” (April) on LinkedIn

MORE

  • “The Last Attention Grab” (April 29) – Teads CVP of Product, Bostjan Spetic, on his personal blog
  • Marketers question expensive AI visibility tools as inconsistent results fuel skepticism (May 1) – Digiday (subscription)
  • Opinion: “Your agency told you they do AI search. Ask them to prove it” (May 1) – James Richardson, consultant in Australia’s AdNews
  • How Integrity Marketing’s AI is creating ‘super advisors’ (May 1) – Insurance NewsNet