There are no big names backing the license yet, but Calliope has already lined up several influencer marketing agencies, such as Viral Nation, to attract clients. “I’ve been getting really good feedback from creators,” says Bianca Serafini, head of content licensing at Viral Nation. She’s confident that a large number of the company’s client list — which is nearly 900 YouTube users — will participate. “No one has presented us with anything like this before.”
In fact, Calliope joins at least one other startup in offering this type of license. Avail launched its own AI licensing initiative called Corpus earlier this year. In late August, the AI startup announced its own partnership with a talent agency representing YouTubers, including popular DIY account JerryRigEverything.
And what does YouTube make of all this? Davis hasn’t worked directly with the company on this project, but believes it’s in line with the video giant’s wishes. “My view is that YouTube wants to give creators more control,” Davis says.
While YouTube won’t comment on specific licensing companies, it does encourage its users to make their own agreements. “Broadly speaking, creators can make deals with third-party companies regarding their content on our platform,” said YouTube spokesman Jack Mallon, who noted that the company recently published a blog post outlining its intentions to allow YouTube users “more control” in the age of AI. The key for YouTube is authorization or obtaining express permission: “Unauthorized access to creator content is prohibited by YouTube’s Terms of Service, and we will continue to implement measures to ensure that third parties comply with these terms.”
Whether the License to Scrape program succeeds will depend on more than securing famous YouTubers. This will require a major shift in the way AI companies approach basic training. With more than 30 copyright cases involving data depletion without permission making their way through US courts, this type of change could prove legally binding. However, since text-to-video generation tools often need large amounts of high-quality data to work well, looking for more sources of said data may require a different approach.
Meanwhile, supporters of this project believe it’s time to move forward rather than wait for AI companies to signal interest. “We just have to get ahead of it,” says Serafini.
Update 10/2/24 9:25 a.m. ET: This story has been updated to include startup Avail, which offers a similar licensing plan.
