Google DeepMind CEO ‘Surprised’ OpenAI Is Moving So Fast on Ads in ChatGPT

Google DeepMind CEO Demis Hassabis says he’s “surprised” by how quickly OpenAI is moving to introduce ads inside ChatGPT  a decision he suggests could complicate how people trust AI assistants.

Speaking with Axios at the World Economic Forum in Davos, Hassabis said monetizing AI through advertising is something Google is approaching “very carefully.” Despite Google’s deep reliance on ads as a business, he emphasized that DeepMind isn’t under pressure to rush into ad-supported AI products.

His comments came shortly after OpenAI announced it would begin testing ads in ChatGPT, tapping into its massive base of roughly 800 million weekly active users  particularly those on the free tier. With rising infrastructure and energy costs, ads may be a practical move for OpenAI, but Hassabis hinted the trade-offs could be significant.

“I’m a little bit surprised they’ve moved so early into that,” he said. While acknowledging that ads have funded much of the modern internet and can be useful when done well  Hassabis questioned how they fit into the role of an AI assistant.

“If you think of a chatbot as something that works for you as an individual, trust becomes essential,” he explained. “So how do ads fit into that model?”

Hassabis reiterated that Google currently has no plans to introduce ads into its AI chatbot, saying the company prefers to observe how users respond elsewhere before making any decisions.

That caution may be well-founded. Consumer backlash has already surfaced around the idea of ads creeping into AI conversations. Last month, OpenAI experimented with suggesting apps during chats a feature users quickly criticized as intrusive. Although OpenAI later clarified that the suggestions weren’t paid ads and disabled the feature, the damage was done.

What upset users wasn’t whether money changed hands, but the feeling that the experience itself had been compromised.

Hassabis echoed that concern, noting that chatbots are fundamentally different from search engines. In search, user intent is clear, making ads easier to integrate without feeling disruptive. AI assistants, however, are designed to be more personal tools that understand users and help manage different aspects of their lives.

“That’s a very different use case,” he said. “It has to be thought through very carefully.”

Google, meanwhile, is leaning into personalization without advertising. The company recently announced new features for Gemini that allow users to opt into personalized responses by connecting Gmail and Photos within Search’s AI Mode. Similar capabilities were also added to the Gemini app, where it can reference a user’s Search and YouTube history.

The challenge, as Hassabis sees it, is that while personalized ads help sustain the free web, inserting ads into an AI conversation can feel jarring even unwelcome. Past attempts by Amazon to add ads to Alexa faced similar resistance, as users wanted a helpful assistant, not a digital salesperson.

For now, Hassabis says DeepMind is taking the long view.

“We don’t feel any immediate pressure to make knee-jerk decisions,” he said. “Our approach has always been to be scientific, rigorous, and thoughtful whether that’s about the technology or the product.”