Amazon Web Services is joining the Deepfake Detection Challenge, a contest started by Facebook in which developers will try to create the most accurate deepfake detection tool.
A deepfake video presents a realistic AI-generated image of a real person saying or doing fictional things.
AWS will contribute cash and technical expertise to the initiative, which was launched in early September. AWS will also sit on the Steering Committee on AI and Media Integrity, which oversees the challenge.
Microsoft and the Partnership on AI (which includes Amazon, Google, DeepMind, and IBM) are also participating, as well as academics from MIT, Oxford, Cornell Tech, UC Berkeley, and others. The tech companies will contribute cash and technology and will help with judging detection tools, a Facebook spokesperson told me.
AWS is also throwing in $1 million to help fund the challenge.
The teams that create the winning deepfake detector will have the option of hosting their models in the AWS model marketplace if they choose.
The group of companies and academics has been creating a benchmark product that can be used by people developing deepfake detection tools to measure the effectiveness of their technology. The benchmark includes two batches of videos for testing—one created by Facebook using paid actors and the other crowdsources from users.
“Competitions bring us together, help us clarify the question to be solved, identify actionable steps we can take, and define a yardstick for progress,” said AWS fellow Pietro Perona, who is also an electrical engineering professor at Caltech. “This competition will unleash the talent of gifted researchers around the world to help maintain a bright line between reality and fiction.”
Facebook CTO Mike Schroepfer said he started the deepfake challenge in part because he remembers being caught off guard by the Russian disinformation campaign on Facebook that helped elect Donald Trump in 2016. He said he doesn’t want to be caught off by deepfakes in 2020.
“This is a constantly evolving problem, much like spam or other adversarial challenges,” Schroepfer wrote in an earlier blog post, “and our hope is that by helping the industry and AI community come together, we can make faster progress.”