Exclusive: DeepMind CEO "surprised" OpenAI moved so fast on ads
Add Axios as your preferred source to
see more of our stories on Google.
Demis Hassabis on stage at Axios House Davos. Photo: Dani Ammann Photography for Axios
DAVOS, Switzerland — As Silicon Valley races to monetize AI chatbots, Google DeepMind CEO Demis Hassabis told Axios he was "a little bit surprised" OpenAI moved so fast on introducing ads in ChatGPT.
Why it matters: Hassabis, a Nobel laureate and one of the world's most influential AI leaders, said Google's Gemini assistant currently has "no plans" to incorporate ads — and cautioned that rushing advertising into AI assistants could undermine user trust.
- OpenAI announced last week that it will begin testing advertising in the U.S. in the coming weeks, as pressure mounts on AI giants to monetize models with soaring development and compute costs.
- The company says ChatGPT's responses won't be influenced by advertisers, but that users' conversations — while remaining private — will shape the ads they see.
What they're saying: "There's nothing wrong with ads. They funded much of the consumer internet," Hassabis told Ina Fried at Axios House Davos on Wednesday.
- "But if you think of the chatbot as an assistant that's meant to be helpful — ideally the kind of technology that works for you as an individual — then there's a question about how ads fit into that model," he said. "No one's really got a full answer to that yet."
- Hassabis stressed that his team is thinking "very carefully" about ads and will monitor how users respond to OpenAI's changes — but said "we don't feel any immediate pressure to have to make knee-jerk decisions like that."
Between the lines: Hassabis drew a sharp distinction between advertising in AI assistants and Google's AI-powered search ads, arguing that search is driven by clear user intent while assistants are meant to work on users' behalf.
The big picture: Hassabis said there's never been anything like the "ferocious competition" and concentration of talent and resources in the AI race, but that he believes commercial pressures are "actually driving the right type of behavior."
- He argued that enterprise customers, in particular, are pushing AI providers to offer stronger guarantees around safety, security and how systems behave with sensitive data.
- Hassabis said that market pressure could serve as a "training run" for managing the far greater risks that would come with more autonomous, agent-based AI systems.
What to watch: Hassabis warned that institutions, governments and businesses are not ready for the transformative impact of artificial general intelligence.
- The "good news," Hassabis said, is that he believes AGI is still 5-10 years away — longer than timelines suggested by peers at Anthropic and OpenAI, who say it could arrive as early as 2026 or 2027.
- "I think as we get closer and we have more agentic autonomous systems, it will become clearer to more of the world that we need to get our act together," Hassabis predicted.
