The CCDH study's co-authors ran tests on these image generators using 40 text prompts related to the 2024 U.S. presidential election. They found that the generators produced deepfakes in 41% of the tests, despite some having policies against election disinformation. The study also highlighted the role of social media in spreading these deepfakes. The authors suggest that AI tools and platforms need to provide responsible safeguards and that social media platforms need to invest in trust and safety staff dedicated to safeguarding against the use of generative AI to produce disinformation. They also call on policymakers to use existing laws to prevent voter intimidation and disenfranchisement arising from deepfakes and to pursue legislation to make AI products safer by design and transparent.
Key takeaways:
- A study from the Center for Countering Digital Hate (CCDH) found that AI-generated disinformation, particularly deepfake images related to elections, has been increasing by an average of 130% per month on X (formerly Twitter) over the past year.
- The study also revealed that free, easily-jailbroken AI tools and inadequate social media moderation are contributing to a deepfakes crisis, potentially undermining democratic exercises like the U.S. presidential election.
- Despite specific policies against election disinformation, AI image generators like Midjourney, OpenAI’s DALL-E 3, Stability AI’s DreamStudio, and Microsoft’s Image Creator were found to produce deepfakes in nearly half of the tests run by the study's co-authors.
- The CCDH calls for AI tools and platforms to provide responsible safeguards, invest in trust and safety staff, and collaborate with researchers to prevent misuse. They also urge policymakers to use existing laws to prevent voter intimidation from deepfakes and pursue legislation to make AI products safer and more transparent.