According to a report from Bloomberg, Reddit has signed a content licensing deal allowing AI models to “train” on its users’ submitted data to the sum of $60 mil every year. Officially, Reddit has declined to comment on the matter, but the timing of it aligns with expectations of its first Initial Public Offering (IPO) in the stock market in the coming months.
As expected, the move has been met with scrutiny and backlash in the hours since the story first dropped, but it isn’t clear what recourse people reluctant to share their comments with AI engines have, at least outside of some direct legal action. Social media platforms like Reddit monetize user data all the time— however, the long-term viability of generative AI without legal challenges seems questionable. The viability becomes even murkier on a platform like Reddit, where copyrighted or even pirated content is often posted, even if it gets taken down.
Since the purported AI content licensing deal (and Reddit itself) has yet to be made public, there is still a chance that Reddit may decide against implementing it, or the final sum may be significantly different.
However, Reddit’s actions could potentially encourage similar moves from other social media platforms. This would be a major headache to artists and other such users who don’t want their work used to train AI models that many say resemble automated content theft machines more than intelligent entities.
Unfortunately, it seems likely that users on any platform will have no recourse until the law is truly “written” around the use of generative AI, copyrighted works therein, and so on. Major cases like The New York Times against OpenAI will ultimately determine the long-term fate of business arrangements such as this.
According to a previous Washington Post report and an anonymous source, Reddit has previously expressed willingness to cut off search engines from Reddit posts, declaring the service “can survive without search.” That threat was reportedly because Reddit wanted to sell AI training data.