A recent study from Clemson University uncovers a troubling development in political discourse on social media platforms. An intricate network of AI-powered bots has been found to be spreading political propaganda, particularly supporting Republican candidates and former President Donald Trump, on the social media platform X (formerly known as Twitter).

Details of the AI Campaign

The campaign employed large language models (LLMs), the same technology behind sophisticated chatbots like ChatGPT, to interact with users on X. While the operators behind this network remain unidentified, the report suggests that this is likely an American-originated effort, as it targeted domestic politics rather than reflecting foreign interests.

The reported network consisted of 686 verified accounts disseminating over 130,000 posts since early 2023. These accounts focused on key political races, including Senate and primary elections, and predominantly backed Trump’s re-election bid. Following inquiries from NBC News, many accounts were subsequently removed from the platform.

Tactics and Impact

The bots employed images appealing to conservative audiences, such as Pepe the frog memes, patriotic symbols, and religious iconography. They mainly engaged by responding to discussions on political topics, thereby amplifying their reach despite having few direct followers.

The influx of AI-driven accounts highlights a growing concern over the manipulation of public opinion through advanced artificial intelligence, especially as the technology evolves and becomes more accessible. The challenge intensifies as these accounts creatively disguise their automated nature.

Tools and Methods

The bot army leveraged Dolphin, an LLM that circumvents restrictions imposed on platforms like ChatGPT, enabling it to produce content without adhering to OpenAI's prohibitions against misleading use. Notably, these AI accounts frequently generated unique hashtags, a telltale indicator of their artificial origins.

Ethical Considerations and Legal Boundaries

This AI use-case exemplifies the dual-edged nature of technology that can be wielded for both beneficial and deceptive ends. Eric Hartford, creator of Dolphin, supports the notion that tools manifest the intentions of their users and recognizes the inevitability of their misuse for political deception.

Despite the evident risks, regulatory gaps persist, as there are no explicit federal laws demanding bots to self-identify as non-human. Nonetheless, the use of AI for political influence, particularly by domestic actors, presents a complex legal landscape.

Broader Implications

The involvement of AI in political propaganda raises significant concerns ahead of upcoming elections. While the U.S. intelligence community actively counters foreign disinformation, it lacks comparable strategies for domestic manipulations of discourse. Given the evolving capabilities of AI, addressing its ethical and political implications remains a pivotal challenge for regulators and social media platforms alike.

For further reading and access to the original report, see NBC News.