Hackfest 2023 - Back to the Future

Hacking Reality: CounterCloud's AI-Driven Disinformation Campaign
2023-10-14, 11:00–11:50, Track #1

Ever wondered if Skynet could run a fake news empire and win a Pulitzer? This is CounterCloud, a two month online experiment that is part 'Terminator,' part 'Black Mirror,' and takes us deep into the wild frontier of totally autonomous AI-generated disinformation.

This presentation aims to provide an in-depth exploration of CounterCloud, a groundbreaking autonomous AI system designed to generate and disseminate disinformation.

Part 1 | “Liar Liar, Pants on Fire.”

The talk will kick off with an overview of the current landscape of AI in disinformation campaigns, highlighting specific case studies and limitations. At this point, nearly all AI-based disinformation campaigns require human interaction and support. So, if we could remove as many humans from the equation, the AI system could operate much faster. The presentation will then introduce CounterCloud, from its architecture and technical aspects to its ethical implications and potential risks.

Part 2 | “Deus Ex Machina”

The presentation will then dissect how CounterCloud uses large language models like ChatGPT to autonomously scrape, generate, and distribute disinformation at scale. It will cover the system's architecture, including its backend setup, the role of cloud computing, and the gatekeeper module that decides what content to target. The output capabilities of CounterCloud, such as generating fake articles, comments, images, and social media posts, will also be discussed in detail.

Part 3 | “The Wolf at the Door?”

Concluding the talk, the focus will shift to the public’s response to the experiment. First broken to the public in an article from The Debrief, and also covered by Wired, the presentation will shift to the social reaction to, ethical considerations of, and risks associated with autonomous AI disinformation systems like CounterCloud. Whether one thinks this is awesome tech or the end of the world as we know it, this experiment proves mass AI-powered disinformation is possible. So now what? The presentation will emphasize the need for public awareness, regulatory challenges, and potential ethical solutions to mitigate the risks.

Are you releasing a tool? – No