ChatGPT ads switch on in Norway — Free and Go see them today

OpenAI is switching on ChatGPT ads in Norway and the rest of the EEA plus Switzerland today. It is the largest expansion of the ad product since the U.S. test began in February, and the first time Norwegian Free and Go users will see sponsored messages inside the assistant itself.

The company announced the expansion on 18 August. The 24 August launch is not a new model drop. It is a change in how Europe’s most used AI surface is financed. For the CIO, CISO, CMO and the board, the question is the same: what happens when advertising moves into the tool staff already use to write, search and decide?

What OpenAI is actually turning on

OpenAI’s own post says ChatGPT Ads will expand to 31 European countries, including Germany, France, Spain, Italy, Sweden, Norway, Denmark, the Netherlands and Austria. Help-center tables and independent reviews of OpenAI’s country list point to the full EU plus Iceland, Liechtenstein, Norway and Switzerland.

Ads appear only on Free and Go. Plus, Pro, Business, Enterprise and education accounts stay ad-free. Units sit below the answer, carry a sponsored label and are kept visually separate from the model’s text. OpenAI says advertising will not influence answers, and that advertisers will not receive conversations, chat history, memories, names, email addresses, precise location or IP addresses. They get aggregated figures such as views and clicks.

Temporary chats are meant to stay ad-free. Accounts OpenAI judges to belong to someone under 18 are also excluded. Political advertising is closed, and ads are not supposed to appear next to sensitive subjects such as physical health, mental health and politics.

That is tighter than some European coverage of a “consent first” launch. OpenAI’s own FAQ, reported independently on 22 August, says personalized ads are not initially available in the EEA or Switzerland. What decides the ad you see today is the conversation you are having, plus rough location and language. Earlier chats, saved memories and ad history stay out until personalization arrives as a separate step.

The real off switch is therefore not a privacy toggle. It is a plan change. OpenAI offers an ads-free version of Free in exchange for fewer messages and fewer tools, including image generation and deep research. Go users must drop back to Free or step up to Plus if they want the ads gone.

Who can buy, and why OpenAI is in a hurry

Advertisers in the new markets buy first through OpenAI’s Ads Solutions team and agency and technology partners. Digiday reports that Publicis, Omnicom, WPP, Havas, Dentsu and MediaPlus are on the list from Monday. Self-serve Ads Manager is due later this quarter. In OpenAI’s own country table, Ads Manager still read “Coming Soon” for all 31 markets just before switch-on.

Dave Dugan, OpenAI’s vice president of ads, told Digiday that Europe is dynamic precisely because there is no single European audience, and that expectations for relevance, quality and trust are high. He has previously said that around 20 percent of ChatGPT queries have direct commercial intent. This product looks more like a buy on an answer surface than a banner.

For OpenAI, ads are also a revenue line against an infrastructure bill that still outruns sales. That makes European privacy more than legal decoration. If personalization comes later, consent, documentation and possible DSA status will decide whether the product can scale here.

What this means for Norwegian leaders

Norwegian organizations in practice have three ChatGPT patterns: governed Enterprise or Business accounts, licensed Plus/Pro seats, and a large field of Free and Go that staff use on their own. Only the last group sees ads today. That is still where the risk sits.

First, Free and Go become a brand environment. An employee who pastes a customer case, a tender draft or an internal analysis into consumer ChatGPT now sits in a surface where sponsored messages appear under the answer. OpenAI says the ad does not change the answer. The board should still treat this as a trust surface, not a neutral notepad. The question is not whether OpenAI sells the chat. The question is whether the company wants decision support and brand to live in the same window as paid influence.

Second, this is a shadow-IT issue. Many Norwegian organizations already ban or restrict consumer ChatGPT for internal data. Where the ban is porous, the ads become a visible symptom. Where the ban is real, IT should still check whether Free/Go is used on work machines, and whether browser extensions, Mac integrations and messaging plugins give the assistant more access than the user thinks.

Third, the privacy layer is split. At launch the serving is contextual, not full profiling. That is better than U.S. personalization. It is not the same as “no processing”. Rough location, language and conversation topic are still signals. When OpenAI later asks for consent to personalize, the organization needs to know whether staff are answering as private individuals inside a work surface.

Fourth, the CMO map changes. ChatGPT becomes a new intent channel in Norway from today, but not a self-serve mass market. Buying goes through selected agencies. Measurement is immature. Politics and health are closed. Anyone who books without an internal rule for AI answer surfaces risks buying space next to their own customers’ questions about defects, prices or complaints.

Brussels may raise the bar

The European Commission is assessing whether ChatGPT Search should be designated a very large online search engine under the Digital Services Act. The threshold is 45 million monthly users in the EU. European reporting says OpenAI reported about 159.1 million monthly recipients of ChatGPT Search in the EU in the six months to 31 March 2026. Commission spokesman Thomas Regnier has said such a designation is “definitely possible”.

The DSA already requires clear ad labelling, a ban on targeting based on sensitive personal data, and a ban on targeted advertising to minors. VLOP or VLOSE status adds systemic-risk assessments, independent audits, data access for regulators and researchers, and a public ad repository. For Norwegian boards the point is simple: the ad surface may become more tightly regulated than the chatbot was last year.

Decisions this week

• Map where ChatGPT is actually used. Separate Enterprise, Plus/Pro and Free/Go. Count shadow use, not just licences.
• Make Free and Go an explicit no for internal and customer data, or give people a governed, ad-free surface. Do not leave the plan choice to each employee.
• Update the privacy and vendor assessment. OpenAI flagged European privacy terms in June and posted an updated EU policy on 14 August. Legal basis, later consent and up to 30 days of retained ads data belong in the DPIA, not a footnote.
• Give the CMO and procurement a shared rule for buying on AI answer surfaces: which brands, which topics, which measurement, and what stays off limits until self-serve and audit trails exist.
• Watch the DSA track. If ChatGPT Search is designated a very large search engine, both duties and documentation change for everyone who uses or buys on the surface.

ChatGPT ads in Norway are not a small product tweak. This is the first time a frontier lab is putting paid attention inside the assistant Norwegian staff already treat as a colleague. Enterprise does not see the ads. Shadow use still does.

Sources and media

• Primary source: OpenAI, “ChatGPT Ads expands across Europe”, 18 August 2026: https://openai.com/index/chatgpt-ads-expands-across-europe/
• Digiday, “OpenAI’s ads business hits Europe at the six month mark”, 19 August 2026: https://digiday.com/marketing/openais-ads-business-hits-europe-at-the-six-month-mark/
• Notebookcheck, “ChatGPT ads hit Europe on Monday, but not the personalized kind”, 22 August 2026: https://www.notebookcheck.net/ChatGPT-ads-hit-Europe-on-Monday-but-not-the-personalized-kind.1375456.0.html
• EU Perspectives, “ChatGPT ads enter Europe as Brussels weighs tougher oversight”, 20 August 2026: https://euperspectives.eu/2026/08/chatgpt-ads-enter-europe-eu-scrutiny/
• Thumbnail: OpenAI Image 2 / hogby.ai