PAI and First Draft Launch Research Fellowship to Combat Media Manipulation

Partnership on AI and First Draft Launch Research Fellowship to Combat Video Manipulation

In an era where artificial intelligence is making media manipulation increasingly sophisticated, a new initiative aims to address the growing threat of deceptive videos. The Partnership on AI (PAI) and First Draft, a non-profit organization focused on combating global disinformation, have announced a joint Research Fellowship dedicated to investigating effective methods of communicating video manipulations to the public.

As we approach 2020, the rise of manipulated and synthesized videos presents a significant threat not only to election integrity worldwide but also to the fundamental concept of truth in media. With technology advancing rapidly, fake videos are becoming increasingly difficult to identify with the naked eye, creating an urgent need for solutions.

“It is vital to consider how the media ecosystem and technology platforms can prevent audiences from being misguided by deceptive information,” said Claire Leibowicz, Founder and Co-Director, who leads research at the Partnership on AI. The fellowship will be situated within PAI’s AI and Media Research focus area, which examines the emerging threat of AI-generated misinformation and its impact on public discourse.

While AI technologies have created new avenues for video synthesis, the fellowship will address the full spectrum of video manipulations, including non-AI techniques. This comprehensive approach recognizes that deception comes in many forms, from sophisticated deepfakes to simpler manipulations like the viral video of Nancy Pelosi from May 2019, which was slowed down to make her appear intoxicated.

The incident with the Pelosi video prompted PAI, the BBC, and WITNESS to co-host a workshop focused on misinformation, bringing together technology companies and journalistic entities to discuss response strategies for manipulated content. During this workshop, participants created an initial typology of possible video manipulations and explored ways to effectively communicate these techniques to the public.

Research indicates that visual information tends to be more memorable than text, which has significant implications for how manipulated or synthesized videos should be labeled. “While many responses to misinformation derive from research on text-based content, there is also a need to investigate how video and photo content should be treated to thwart the spread of misinformation,” Leibowicz explained.

One challenge in this field is that much of the existing research has been conducted in controlled academic environments rather than tested in real-world scenarios where information flows freely across platforms. The fellowship aims to address this gap by developing universal visual cues—similar to Twitter’s blue verification badge or the hashtag—that could effectively signal manipulated content across different platforms and websites.

Key questions the fellowship will explore include whether descriptive labels on videos that explain manipulation techniques could educate viewers about deceptive practices, and how platforms and newsrooms might collaborate on typologies and strategies for labeling manipulated content.

First Draft brings valuable experience to this partnership, having worked with news and civil society partners worldwide since 2015 to test different methods for combating misinformation around elections. The organization has specifically explored the challenge of labeling manipulated media, recognizing that this complex issue requires a collaborative approach.

The fellowship is grounded in the belief that work on media manipulation should draw from a diverse range of stakeholders, including civil society organizations, technology companies, media entities, and academic institutions. PAI and First Draft are seeking candidates from across these sectors, particularly those with backgrounds in psychology, communication, political science, human-computer interaction, or other related disciplines.

As misinformation continues to evolve beyond simple “fake news” to include sophisticated visual deception, this initiative represents an important step toward developing standardized, research-based approaches to help audiences navigate an increasingly complex media landscape.