A week in Generative AI: Settlements, Hiring & Dishes
News for the week ending 7th September 2025
It was Labor Day weekend in the US which inevitably means a quieter week on the AI-front. We had a big copyright ruling in the US though, leading to Anthropic settling with an unprecedented large $1.5bn with content-creators. OpenAI also announced they were building an AI-powered hiring platform, which is a bit weird.
Two great long reads that are Web 4.0 related as well - both Charlie Guo’s piece on how ChatGPT has won the chatbot wars and Simon Willison’s piece on using GPT-5 Thinking for search have some great stats and examples of how generative AI platforms are disrupting the internet.
Enjoy!
Anthropic to pay $1.5 billion to authors in landmark AI settlement
The agreement made this week for Anthropic to pay $1.5bn to authors is a huge deal for so many reasons. It’s the largest publicly reported recovery in the history of US copyright litigation and means that the AI company will be paying authors approximately $3,000 per book or work.
The ruling also means there is now a precedent in the US courts that material used in training data doesn’t necessarily fall under fair use, which could lead to similar ruling against other AI companies. This is obviously a huge win for authors and content owners and we’re likely to see a growing number of paid partnerships for content as a result of this.
It doesn’t come at a bad time for Anthropic who also completed a new funding raise this week, raising $13bn at a $183bn valuation. So they have the cash to settle. But as M.G. Sielgler wrote in A Cynical Read on Anthropic's Book Settlement there could be an unintended consequence on this being such a high settlement figure - a much higher barrier to entry for new competitors. And broadly, less competition is generally a bad thing.
But, this is a huge win for content creators and opens the door to more wins in the future.
OpenAI announces AI-powered hiring platform to take on LinkedIn
This is a bit of a weird one - OpenAI doesn’t usually announce its plans in advance and this new take on AI-powered hiring isn’t due to launch until mid-next year. It’s also an area that OpenAI doesn’t have any experience in, but that certainly hasn’t stopped them before.
The announcement was part of a wider plan on how Fidji Simo, the new CEO of Applications at OpenAI wants to expand the economic opportunities with AI. The hiring platform is a part of this plan, alongside an ambition to certify the AI skills of 10m Americans by 2030 through the OpenAI Academy.
To me it currently feels like OpenAI want to rebuild the entire internet around AI, and specifically around their AI, ChatGPT. Sam Altman has talked before about “Sign in with ChatGPT”, allowing users to take their own personal AI with them around the internet. In the last 12 months they’ve launched ChatGPT Search, ChatGPT Shopping, and there are rumours that they also have designs on building a ChatGPT-powered browser and even a ChatGPT-powered social network. Oh, and the new hardware device they’re now building with Jonny Ive.
From the outside it looks like they’re taking on Google, Amazon, Meta, Microsoft, and Apple all at once in a big play to rule them all. It’s certainly ambitious, and I do think that it’s inevitable that many of these services and technologies get rebuilt around AI from the ground up. But the idea that, if OpenAI are successful, there could be one company who owns/dominates all of these technologies doesn’t sit well. That might be because I’m currently watching the fantastic Alien: Earth though, where Governments no longer exist and five companies control the planet 🤓.
Helix does the dishes
I mean it’s maddeningly slow and that’s not how I’d pack the dishwasher, but it can do it and that means if I owned one then I wouldn’t have to. So a win!
Web 4.0
Two great long reads that relate to Web 4.0 this week. Some great stats about the growth of ChatGPT vs. other chat platforms in The Chatbot Wars Are Over and Simon Willison does a great job talking through how good a replacement for Search GPT-5 is.
Ignorance.ai - The Chatbot Wars Are Over
Simon Willison - GPT-5 Thinking in ChatGPT is shockingly good at search
AI Ethics News
OpenAI to route sensitive conversations to GPT-5, introduce parental controls
Google Gemini dubbed ‘high risk’ for kids and teens in new safety assessment
Attorneys general warn OpenAI ‘harm to children will not be tolerated’
Chatbots can be manipulated through flattery and peer pressure
Warner Bros. Discovery sues Midjourney for generating ‘countless’ copies of its characters
“The future is already here, it’s just not evenly distributed.“
William Gibson