From Wired
Deepfakes Aren’t Very Good. Nor Are the Tools to Detect Them
By Will Knight
WE’RE LUCKY THAT deepfake videos aren’t a big problem yet. The best deepfake detector to emerge from a major Facebook-led effort to combat the altered videos would only catch about two-thirds of them...
Creating a really useful deepfake detector might be even harder than the contest suggests, according to Farid of UC Berkeley, because new techniques are rapidly emerging, and a malicious deepfake maker might work hard to outwit a particular detector.
Farid questions the value of such a project when Facebook seems reluctant to police the content that users upload. “When Mark Zuckerberg says we are not the arbiters of truth, why are we doing this?” he asks.
Even if Facebook’s policy were to change, Farid says the social media company has more pressing misinformation challenges. “While deepfakes are an emerging threat, I would encourage us not to get too distracted by them,” says Farid. “We don’t need them yet. The simple stuff works.”
Hany Farid is a Professor in the School of Information and the department of Electrical Engineering and Computer Sciences at UC Berkeley.
