The AI platform updates agencies need on their radar   

ChatGPT starts selling ads for real  

OpenAI is expanding ‘ChatGPT Ads’ to more than 40 countries following its US pilot, with advertisers initially accessing the platform through OpenAI, agency and technology partners. Self-service access via ‘Ads Manager’ has now launched across Europe, including the UK, as well as India, the Middle East and North Africa.

OpenAI maintains that ads stay separate from its AI-generated answers and advertisers don't receive access to users' private conversations. The platform has moved beyond CPM/CPC to conversion optimisation, with CPC and outcome-optimised bidding now accounting for the majority of campaigns, alongside geo-targeting, custom audiences and measurement via the ‘OpenAI Pixel’ and Conversions API. Product feeds and platform targeting have also been added. Dentsu, Omnicom, Publicis and WPP are already partners, while OpenAI says its ecosystem has expanded to more than 50 technology and measurement partners.

And this is no longer a small-scale experiment. OpenAI says ChatGPT Ads has reached a $1bn annualised revenue run rate in less than 200 days, with tens of thousands of advertisers now using the platform. The company says small and medium-sized businesses now represent a material share of the business, following the introduction of Ads Manager in May.

According to Digiday, six live OpenAI job postings covering growth, data science and sales ops also point to a dedicated SMB advertising push in the works, mirroring the Google and Meta playbook.

OpenAI says its ads system uses the context of the current conversation to show an ad relevant to what someone is exploring. Depending on the country and a user's settings, it may also use context from their broader ChatGPT experience, including interests and preferences.

Reno Hughes, Paid Media Lead at The MTM Agency, argues this "isn't simply another place to buy media." It reflects a shift in how people discover and decide. Rather than "who is this person and what have they done before?", the relevant question becomes "what are they trying to solve right now?" With formats and measurement still maturing, Hughes’ advice holds … there's a case for experimenting now to build fluency, rather than committing serious budget just yet.

Google embeds ads directly inside AI Mode 

Google is moving beyond placing ads around AI search results to building formats inside the conversation. These include "Conversational Discovery" ads, "Highlighted Answers", Gemini-powered "Shopping" ads with contextual product guidance, and a "Business Agent for Leads" letting consumers interact with a Gemini-powered brand agent in-ad rather than filling out a form.  

Google calls it a "new generation of ads for the AI era of Search." These are ads that respond to context rather than simply match a query. It's a meaningful format shift rather than a targeting tweak, and brand teams might want to test how product data translates into a conversational format now, rather than waiting for it to become standard. 

‘Paid Organic’ could create a new way to influence AI 

Sean Betts (Chief AI & Innovation Officer, OMG UK) proposes a third category sitting between paid and organic media. ‘Paid Organic’ means brands paying to feed information into the sources AI platforms draw on, while the AI independently decides whether to use it. "With paid advertising, I pay to appear. With organic, I earn the right to appear. With Paid Organic, I pay for the opportunity to influence how I appear."

Betts points to TIME creating machine-readable articles for AI crawlers, including sponsored information brands can pay to make available. The result is brands "buying consideration, not distribution." 

It's still experimental and not without pushback. Perplexity has reportedly blocked TIME's sponsored content from influencing its models, calling the approach potential "cloaking." Make of that what you will. The open question is disclosure. If sponsored information feeds an AI answer, the "sponsored" label attached to the original content can vanish by the time it's synthesised into a recommendation. 

Meta puts AI agents inside the customer relationship 

Meta's ‘Business Agent’ now lets businesses run personalised customer experiences across WhatsApp and Messenger, with more than one million businesses already using Business Agents. New tools also connect Meta AI directly to ad campaigns, analysing performance, spotting patterns in successful creative and audiences, and recommending budget and optimisation changes. A new ‘Business Agent Platform’ lets companies build and deploy custom agents at scale.  

The line between advertising, customer service and commerce is getting harder to draw, and AI could increasingly handle questions, share product info and guide purchases directly. It is worth mapping which parts of the funnel this quietly absorbs. 

The next shift: from influencing people to influencing agents  

AI is moving from answering questions to taking action, researching, comparing, negotiating and sometimes transacting on a user's behalf. New PHD/WARC research estimates AI-agent-facilitated consumer spending could triple from $944bn (£697bn) this year to $3.35tn (£2.48tn) by 2030, with fastest growth in telecoms, financial services and travel. Zuckerberg has set out a similar vision of personalised AI agents as part of everyday life. The uncomfortable question worth raising with clients now is what happens when the system building the shortlist isn't the consumer anymore?