YouTube has introduced a new weapon to use to combat unauthorized content generated by artificial intelligence, and it gives creators the ability to flag and have removed videos containing their image without permission.
That new weapon, now available to some YouTube Partner Program members, goes a long way toward protecting creators from the new threat of deepfake technology.
The rollout started today, and the initial batch of qualified creators just received the email notification of access to the tool. YouTube will roll the availability out over the next couple of months, rolling ever-increasing numbers of creators into the program.
The new functionality works through the Content Detection section of YouTube Studio, where creators are able to preview videos auto-flagged by YouTube’s computer systems. Following the completion of an identity authentication procedure, creators are able to access a dashboard of possible matches of their own likeness throughout the site.
YouTube Launches AI Tool to Combat Deepfakes
When an author discovers what they think is unauthorized content featuring their face created by an artificial intelligence, they can request removal directly through the interface. The system functions similarly to YouTube’s existing Content ID system, which has successfully identified copyright audio and video content for years.
But YouTube is giving early viewers the heads-up on the current limitations of the technology. In the guide shared with early testers, the company admits the system isn’t even finished yet and “may play videos of your real face, not modified or computer-generated ones.”
That would have creators spotting legitimate clips of their own content alongside potentially concerning deepfakes.
The rollout marks the end of months of trials. YouTube announced the feature initially last year and began a pilot program back in December, and they worked alongside the Creative Artists Agency (CAA) to test the technology among big-name talent.
The firm explained the program as providing “some of the globe’s most influential individuals” access to beta-phase technology built to handle content generated by AI at scale.
The pilot program gave us constructive feedback so the algorithms for detection and the user interface could be tuned prior to today’s wider rollout.
The similarity detection software is just one of the different components of YouTube’s multi-faceted scheme for addressing content developed by algorithms on the platform. In March, the company issued required labeling mandates for uploaders of content developed or modified through the application of AI software.
Can YouTube Foster Innovation While Protecting Creators from Deepfakes?
YouTube has even implemented strict policies related to music made by artificial intelligence, namely content reproducing an artist’s original singing or rapping style. These regulations are the product of the site seeking equilibrium between innovation and safety for creators.
The timing of the safeguards is significant, given that YouTube and parent company Google are busily developing and marketing their own AI-powered video generation and editing software. This simultaneous approach of advancing the push of AI technology and constructing guardrails underscores the tricky place where tech companies sit as generative AI gets increasingly sophisticated.
The rise of deepfake technology has caused much angst not only within the entertainment sector but all the way across the globe. Computer-made videos are capable of appearing believable to replace someone’s face on someone else’s body or make them appear to have uttered something they never did.
For content creators who have built their reputations upon truth and credibility towards consumers, unauthorized deepfake content poses both reputational and financial risks.
Other organizations are taking the threat seriously as well. The Screen Actors Guild (SAG-AFTRA) has been vocal about deepfake concerns, with members like Bryan Cranston speaking publicly about discussions with companies like OpenAI regarding protections for performers.
YouTube’s Drive for Deepfake Detection Accuracy
By expanding access to this detection system, the firm should go on to refine the system even further and eliminate false positives. How much the system performs well would be how well it separates real content and unauthorized deepfakes, a technically challenging task that requires continued development.
For future content creators looking for access, YouTube has failed to provide clear deadlines or eligibility requirements beyond the level of Partner Program membership. Individuals granting access will have the new functionality seamlessly incorporated into the existing YouTube Studio workflow they utilize, making it a seamless addition to their content creator toolbox.
This announcement is an important acknowledgment by YouTube to the reality of creator identity security in the age of AI as not just a nice-to-have but ever more ever-more-necessary infrastructure to the future of supporting digital sites of trust and authenticity.
