@firstdraftpro: Here’s where we’re at (you need to lock in for this one): Early on in the development of First Draft Pro, AI was everywhere. It was clear that because AI was so new, and because there were no guardrails, that legal and ethical lines were gleefully bounded over in its creation. But, it was also clear that the metaphorical cat had clawed its way out of the bag, and the bag was in shreds. There was no going back. I wanted to try to assess what was valuable and useful in this new landscape, while protecting the human-essential role of storytelling. I remember reading a Threads post in 2023 where someone said: “I don’t want AI to love my children for me, I want it to do my dishes”. My early thinking about whether there was an appropriate place for AI in a writing app was along those lines: What’s the equivalent of doing the dishes for a writer, and is there a way that we can use AI to do that? Was there a use for AI that was not creative? The feature I decided to build was one that updated your outline in line with any changes you’d made to your manuscript. Importantly, it was: 1. non-creative, 2. clearly labeled, 3. optional. I even built an all-off switch that would remove it from the interface in order to be respectful to writers who wanted absolutely nothing to do with it while I explored the question. There were still ethical and environmental concerns for me. No reasonable person can dispute the fact that AI was made with stolen work. And that theft makes very clear the value that commercial AI companies put on writers and artists. There are two things that are important for training an AI: data and processing power. Nvidia’s stock soared 300% in the first 12 months of the AI boom. Why? Because they paid for the processing power. But, they stole the data (and had the gall to claim they wouldn’t have been able to make the tech if they had to pay for it). I watched as authors sued for the theft of their work, and was optimistic that courts would take the opportunity to establish guardrails, and set right the wrong that was done when this material was stolen. I (naively, in hindsight) believed that the legal systems would help to define the grey areas in fair use to prevent what is an obvious infringement. That copyright law would be clarified, fair compensation would be given, and that consent would be mandatory going forward. So, what’s changed? Two of the copyright infringement cases levelled against two of the biggest players in the commercial LLM space had judgements that came down. And neither of those cases were ruled in favour of the authors. Sigh. All is not lost. It’s just going to take much longer to establish guardrails. But that is FRUSTRATING, because this was the first opportunity to start to set things right. These cases have been in progress for 2 years, and now it’s going to take longer and be far more expensive for authors to fight for their rights. I was listening to this episode of Scriptnotes where they talk about the concept of having to “turn it in before it’s done”. You have to stand on stage and say a joke to see if it’s funny. Making an app is a lot like that. One of the hard things about building something in public is that you have to put something in the world that’s in-progress, when you’re still exploring what something means and if it’s even useful. And sometimes that means rolling things back once you decide where you stand. When we implemented this feature in First Draft Pro, we got a ton of very positive feedback. That’s why it’s taken so long for me to make this call, despite feeling very uncomfortable about it for a long time. I still think that ✨non-creative✨ AI features could have a place in a writing app, but until there’s an ethical alternative, or commercial AI companies begin to respect author copyright, we won’t be building them anymore. #BookTok #writertok #writersoftiktok #authortok #writingcommunity

Tarryn from First Draft Pro

Open In TikTok:

Region: NL

Thursday 24 July 2025 13:58:57 GMT