OpenAI released images of comics generated by operatives trying to run influence operations. (OpenAI)
ChatGPT’s parent company, OpenAI, is accusing a Chinese technology company of leveraging its signature platform to manipulate the opinions of people who live in the United States, particularly when it comes to two of today’s hot-button issues.
Big picture view:
OpenAI divided the accounts allegedly being used for these campaigns into two groups: one that focused on building opposition to U.S. data centers and that targeted overseas Chinese individuals, and one that took aim at U.S. tariffs and technological dominance.
The Chinese technology company was working for Chinese provincial-level government clients, which is in line with a commercial ecosystem that can be leveraged to promote the priorities of China’s Communist Party.
What they're saying:
Pointing out that discrediting OpenAI was apparently one of the latter group’s goals, the AI giant used the chance to throw some shade at artificial intelligence models being developed in China, writing, "It is ironic that the two operations used American AI, rather than Chinese models, to generate their content about American AI."
How they did it
OpenAI’s statement highlighted the data center campaign to show how influence operators tried to manipulate people. To convince the public, OpenAI said, they would amplify concerns people already had about energy prices and local impacts.
According to OpenAI, the operatives, whose accounts are now banned, would often ask ChatGPT to create English-language comments and images that argued data centers and AI applications were driving electrical bills higher and increasing prices in America. The report included an example from one of the times they asked for comic strips about power grid capacity auction prices.
OpenAI released images of comics generated by operatives trying to run influence operations. (OpenAI)
The company’s statement noted that attempts by operatives to influence people in other nations often rely on aligning themselves with local issues or sincerely-held beliefs to build their own credibility. They would then leverage that trust to try and sway the debate, without letting on who they actually were or what their motivations might be.
The other campaign
The other group identified by OpenAI took aim at the tariffs being imposed by the Trump administration and criticized them as ways for the United States to solidify its technological advantages.
One indicator identified in the report as to who may have been behind this cohort was the prompts, which directly stated that Chinese President Xi Jinping should not be included in the output and that, instead, the responses should only target President Donald Trump.
OpenAI released images of comics generated by operatives trying to run influence operations. (OpenAI)
The ChatGPT accounts in this group were connected to social media accounts that OpenAI said were likely not authentic. The company noted that those accounts also kept OpenAI in their crosshairs by claiming that ChatGPT data was compromised.
What's next:
OpenAI concluded by explaining its choice to release this report on the banned accounts and the suspected links between them, including that they were likely part of a Chinese influence campaign.
The company said it did so to help others in the AI community, governments, and the general public understand and identify the ways artificial intelligence can be used against them by "foreign threat actors to manipulate legitimate public debates, weaken democratic institutions and advance totalitarianism with AI characteristics - the use of AI for surveillance, censorship and control over political, social and private life."
The Source: Information for this article was taken from OpenAI. This story was reported from Orlando.