ChatGPT Ads Goes Self-Service in Belgium: A New Advertising Channel Alongside Google and Meta

Since 31 August 2026, Belgian businesses can create and run their own ChatGPT Ads campaigns, without going through an agency or a technology partner. One week after this channel opened in self-service across 31 European markets, advertising has officially entered conversations with the world's most used AI assistant. Which raises the question: does this new channel already deserve a place in your advertising budget, alongside Google and Meta?

1. What happened in three weeks

On 18 August 2026, OpenAI announced the expansion of ChatGPT Ads to 31 European markets, including Belgium, France, Germany and the Netherlands. On 24 August, ads went live in Belgium, but access remained limited: only OpenAI's sales team, agency partners and technology partners could create campaigns. On 31 August, self-service access opened through the Ads Manager tool across the 31 announced markets: any Belgian business can now sign up at ads.openai.com, verify its identity and launch its own campaigns, much as it would on Google Ads or Meta Ads Manager.

2. A different targeting logic than Google or Meta

On Google, a user types a precise keyword and the advertiser bids on it. On ChatGPT, the user explains what they need in their own words, sometimes across several messages: planning a trip, choosing software, comparing options. That phrasing often reveals far more about a real intent than a single keyword, and it is this understanding of the need that ad targeting relies on. In practice, an ad can appear below a ChatGPT answer, as a separate card clearly labelled "Sponsored". It is only shown to Free and Go accounts: Plus, Pro and Enterprise subscriptions remain ad-free. No ads are shown to accounts identified as belonging to minors, nor within temporary chats.

3. GDPR: personalisation requires prior consent

On 14 August 2026, OpenAI published a specific framework for the European Union. In Belgium, as elsewhere in the EU, using conversation history or memory to personalise an ad requires the user's explicit, prior consent, given in their account settings. Without that consent, the user still sees ads, but based only on the topic of the current conversation and an approximate location. This prior-consent principle differs from the approach initially tested in the United States, where personalisation was switched off after the fact rather than opted into beforehand.

4. Is it already worth testing this channel?

Six months after the pilot launch in the United States in February 2026, OpenAI reports that tens of thousands of advertisers worldwide have already used ChatGPT Ads. The platform is still young: geo-targeting and custom audiences were only just introduced, as was conversion measurement through the OpenAI Pixel and a Conversions API. For a Belgian SME, the opening of self-service mainly lowers the cost of testing: a modest budget is enough to check whether your target personas show up in this kind of conversation, before competitors get there first.

The AI view, humans first

An AI tool can generate ad creative variants, adjust bids continuously, or monitor this new channel's performance hour by hour. But deciding whether ChatGPT Ads makes sense for your business, what budget to give it against Google and Meta, and how to align the message with the rest of your media strategy remain human decisions. AI executes and optimises. Expertise decides where, when and why to invest.