Alex Bestall, CEO of music copyright management firm Rightsify, isn’t surprised if it’s the case that OpenAI wanted to focus on bigger fish. “Startups are much more flexible, but big labs have minimum data volumes to consider any deals,” he says.
Now, OpenAI is facing the prospect of its first copyright infringement litigation in the UK. In addition to its copyright claims, Mumsnet is claiming a breach of its terms of use and alleging database right infringement, meaning the extraction of all or a large part of a database without the owner’s consent.
Mumsnet sent its initial letter announcing it was considering legal action in July. More recently, it received a response from OpenAI with a list of questions. “They did not deny the fact that they had scraped,” she says. As of now, Mumsnet plans to continue on the litigation track; it has not yet determined whether it will file suit in the UK’s High Court or a specialized intellectual property court. (OpenAI acknowledged to WIRED that it had received and responded to the Mumsnet complaint, but did not offer comment on Mumsnet’s legal claims.)
In the meantime, Mumsnet is actively pursuing licensing arrangements with other AI companies. Roberts says that it is speaking with Google, as well as intermediary startups that have cropped up to facilitate data licensing. (Google did not respond to WIRED’s request to confirm these talks.)
“I’m quite worried about the ecosystem, where these big LLMs are allowed to march all over small publishers to build their models, and then people have less reasons to go and visit the websites,” Roberts says. “We need to come to some sort of satisfactory arrangement where people are compensated for their work.”
As Mumsnet’s content is largely user-generated, WIRED asked whether it was considering any sort of payment system for users when it does strike deals. Roberts says there is no plan at the moment, but that she would consider it if data licensing for AI became incredibly lucrative down the road.
She says that, based on comments she received after the announcement Mumsnet was looking into legal action, users by and large understand the company’s aims in licensing their data. “We’re quite concerned about AI being gender-biased,” she says. “There’s something to be said for it being trained on verified female voices.”
Roberts is optimistic about how Mumsnet’s potential legal action will unfold. “We think we have a good chance,” she says. In the US, there have already been dozens of copyright-infringement cases brought against AI companies. In many of the ongoing cases, AI companies are defending themselves by arguing that their actions are shielded by the “fair use” doctrine, which allows for copyright infringement in certain circumstances. The UK has a similar concept, which it calls “fair dealing,” but it’s significantly more limited in scope.
Regardless of the outcome, Roberts is glad her platform is taking a stance. “This is probably more about the principle of the thing than anything else,” she says.
