Political Deepfakes Outpace Cyber Attacks In AI Misuse: DeepMind Study

A recent study by DeepMind, a division of Google’s parent company Alphabet, revealed that the most common misuse of artificial intelligence is the creation of deepfakes featuring politicians and celebrities, rather than AI-assisted cyber attacks.

This misuse is nearly twice as prevalent as the next most common misuse, which involves falsifying information using text-based tools.

The research analysed approximately 200 incidents of misuse from social media and online reports between January 2023 and March 2024.

The study also found that the primary goal of those misusing generative AI is to shape or influence public opinion, accounting for 27% of cases.

Lead author and Google DeepMind researcher Nahema Marchal noted that while there are concerns about sophisticated cyber attacks, the more frequent misuse involves deepfakes, which often go unnoticed.