OpenAI reports China-linked influence campaign on AI data centres

OpenAI has accused China-based actors of using ChatGPT accounts to fuel public opposition to artificial intelligence data centres, marking the latest escalation in concerns about foreign interference in the AI sector.

The San Francisco-based company said it detected and shut down multiple accounts that attempted to ‘exploit and amplify existing public concerns’ about rising energy prices and the environmental impact of AI infrastructure. The campaign targeted communities in the United States and Europe where new data centres have been proposed.

Coordinated Messaging Across Platforms

According to OpenAI’s threat intelligence team, the accounts engaged in coordinated behaviour between January and March 2024. They posted content that emphasised the energy consumption of AI facilities, with some claiming that a single data centre could use as much electricity as 50,000 homes. The accounts also shared links to local news articles and petition websites opposing data centre construction.

But the messaging didn’t appear organic. OpenAI identified patterns suggesting the accounts were part of a directed campaign rather than genuine grassroots activism.

The company found approximately 130 ChatGPT accounts linked to the operation, though the actual reach of the campaign remained limited. Most posts generated fewer than 100 engagements on social media platforms where they were shared.

Exploiting Real Concerns

Energy consumption by AI data centres has become a legitimate flashpoint in communities worldwide. Microsoft’s plans for a facility in Wisconsin faced opposition last year, whilst similar projects in Ireland and the Netherlands have sparked protests over grid capacity.

That’s what makes this type of influence operation particularly insidious. It takes genuine local grievances and weaponises them.

An OpenAI spokesperson stated: ‘We’re committed to detecting and disrupting deceptive use of our platform. Whilst concerns about AI infrastructure are valid and deserve serious discussion, coordinated inauthentic behaviour undermines that democratic process.’

Growing Scrutiny of AI Infrastructure

The disclosure comes as Western governments race to build AI capabilities whilst China pursues its own aggressive expansion. Data centres supporting large language models require enormous amounts of electricity, with some estimates suggesting AI could account for 3-4% of total US power demand by 2030.

OpenAI said it shared its findings with law enforcement and other technology companies. Meta and Google have previously reported similar influence operations targeting technology policy debates, though this appears to be the first documented case specifically focused on AI infrastructure opposition.

So far, Chinese officials haven’t responded to the allegations. But the incident raises questions about how foreign actors might shape public opinion on critical technology decisions in the years ahead.