OpenAI claims China-linked groups used ChatGPT for US influence ops

OpenAI has claimed that forces suspected of being linked to China attempted to manipulate public opinion in the United States using ChatGPT.

In its 'Threat Report' released on the 10th (local time), OpenAI announced that it had detected and blocked two clusters of ChatGPT accounts suspected of being linked to China.

According to the report, these groups were active from late last year to early this year, and they attempted to create division by delving into issues related to artificial intelligence (AI) policy in the United States, including energy issues and trade policies such as tariffs.

One group led a campaign to incite that AI data centers are increasing electricity bills for ordinary households.

They requested ChatGPT to draw cartoons about electricity market auction prices based on media reports, or to add text to general electricity market images such as 'The AI industry is booming, but the cost is borne by ordinary people.'

Although they disguised their access as if from outside China using VPNs, the commands (prompts) entered into ChatGPT were in simplified Chinese used in mainland China.

The report analyzed that they were likely private companies in China with provincial governments as clients.

Another group waged a campaign criticizing US tariff policy as an attempt to seize technological hegemony.

They mass-generated political satire cartoons on the theme of US-China technological competition through ChatGPT, and it was found that they consistently entered commands such as 'The cartoon should only feature President Donald Trump, not Chinese President Xi Jinping.'

The generated cartoons were distributed through social networking services (SNS) such as X (formerly Twitter).

It also included destructive content where President Trump wielded a hammer at a wall labeled 'Future of the World' or sawed the ladder he was climbing.

The report added that they referred to themselves as 'suijun' (水軍), which is Chinese slang for online accounts engaged in organized criticism activities.

However, the public opinion manipulation attempts by these groups did not exert significant influence for the most part.

Regarding OpenAI's announcement, the Chinese Embassy in the US countered, stating, 'We firmly oppose baseless attacks or slander against China,' and 'The Chinese government is working to ensure that AI becomes a force for good for all,' according to Reuters.