Adobe, Twitter, NYT launch effort to fight deepfakes
Illustration: Sarah Grillo/Axios
Hoping to stem a forecast rising tide of faked video, Adobe, Twitter and the New York Times are proposing a new industry effort designed to make clear who created a photo or video and what changes have been made.
Why it matters: With editing tools and artificial intelligence rapidly improving, it will soon be possible to make convincing videos showing anyone saying anything and photos of things that never happened.
How it works: Adobe is proposing an opt-in system allowing creators and publishers a secure way to attach attribution data to content. Adobe could include the technology in its tools, but it would be an open standard that others could use as well. Adobe is showing a prototype today at its MAX conference in Los Angeles.
The big picture: Adobe and Twitter are not alone in seeing authentication as a key to fighting deepfakes. Among the players in the space is a startup called Truepic, which aims to create a secure path from the moment a photo or video is captured.
Thought bubbles from Axios' Kaveh Waddell:
- This solves a small but important layer of the online trust crisis. This would allow a reader to verify that something came from Axios — but if they are skeptical of Axios to begin with, that won't matter.
- Verification that isn't easily accessible threatens to bifurcate online information into "trusted content" from those who have the resources to verify it and an easily dismissed information underclass.
The big question: How are these companies planning to prove authenticity?
- Most early attempts have leaned on blockchains, or decentralized lists of transactions that can't be altered. These can increase trust, but are harder to interact with.
- The alternative is a database held by a single company — who would technically have the ability to change the entries.
- Adobe says it hasn't finalized what mechanism it will use, saying it wants its partners to have a say.
What they're saying:
- Adobe general counsel Dana Rao: "When it comes to the problem of deepfakes, we think the answer really is around 'knowledge is power' and transparency. We feel if we give people information about who and what to trust, we think they will have the ability to make good choices."
- New York Times' head of R&D Marc Lavallee: “Discerning trusted news on the internet is one of the biggest challenges news consumers face today. Combating misinformation will require the entire ecosystem — creators, publishers and platforms — to work together."
- Twitter trust and safety head Del Harvey: “Serving and enhancing global public conversation is our core mission at Twitter. Everyone has a role to play in information quality and media literacy."
What's next: Adobe plans a summit next month at its headquarters for all interested parties. "We do look at this as a shared responsibility," Rao said.