Reflections on Mumsnet and the Challenges of AI
When I launched Mumsnet in 2000, the forum was primarily a space where I engaged in conversations with myself through various personas. Little did I know that nearly 25 years later, millions of users would contribute over six billion words, discussing topics ranging from mundane parking dilemmas to complex geopolitics, and even the infamous “penis beaker” discussion—an exploration of one user’s post-coital clean-up routine that attracted so much traffic it nearly brought our website to a standstill for two weeks. Today, those billions of words are central to one of the pressing issues of our time: how do we reconcile the advancements in AI with the need for copyright protections for publishers?
Last year, Mumsnet took a significant step by becoming the first company in the UK to initiate legal proceedings against the tech giant OpenAI, the creators of ChatGPT. We discovered that they were scraping our site in order to train their large language model (LLM), which we believe constitutes a violation of both our terms of use and copyright law. When we proposed that they consider licensing our content, they initially hesitated and eventually expressed a preference for “less open” sources of data. The truth is, they had already harvested our content—something they’ve not denied in any of our communications.
Ironically, we would welcome collaboration with companies like OpenAI. We believe that the female-centric conversations on Mumsnet could aid in addressing the misogyny frequently embedded in most AI models. In fact, we utilize OpenAI’s LLM in our own MumsGPT—a question-answering tool that provides valuable insights derived from the discussions among Mumsnetters. However, the key difference is that we paid for and licensed the technology we use.
Mumsnet is uniquely positioned to withstand the repercussions of AI chatbots. A significant portion of our traffic comes directly to us, and while an LLM may produce a Mumsnet-style response, it cannot replicate the rich, empathetic, and often humorous interactions that occur on our platform. When parents face sleepless nights with their babies, a generic AI response will never compare to the authenticity of advice from someone who has experienced the same struggle at 3 a.m. Additionally, AI lacks the ability to offer the genuine support that assists around a thousand women each year in escaping abusive relationships, thanks to the solidarity of other Mumsnetters.
However, if tech giants are permitted to exploit publishers’ content without repercussions, they risk obliterating many of these publishers. Thus, it is surprising that the current government appears determined to dismantle our long-standing copyright laws. Their proposed changes would require publishers to actively opt-out of having their content scraped, or risk having it taken without compensation or consent—akin to allowing someone to rob your home unless you’ve clearly posted a sign prohibiting it.
- This approach threatens not only those being scraped, but also jeopardizes the integrity of AI models. If it becomes financially unfeasible for publishers to create the content that AI systems rely on, they will eventually cease to exist, leaving nothing for these companies to train their models on.
- The Government has frequently highlighted the “uncertainty” surrounding the UK’s copyright framework in relation to AI. In reality, our robust copyright system is one of our greatest assets. If the government aims to provide clarity, why did it oppose the Kidron amendments introduced in the Data Bill, which sought to adapt the existing copyright framework for the AI age?
AI holds immense potential to propel human progress and enhance our lives, and we certainly want the UK to benefit from this technological evolution. However, that does not warrant publishers giving away their content to AI corporations without compensation. Just as tech firms do not expect the electricity powering their data centers to come free of charge, why should they not also compensate the creators whose content is essential for training their models?
The UK has the opportunity to lead globally in this arena. To achieve this, we must foster a sustainable ecosystem where innovation and fairness coexist harmoniously.
Justine Roberts is the founder and CEO of Mumsnet
