Two huge music publishers just sued Anthropic, accusing them of conducting “a brazen campaign of illegally torrenting, scraping, and downloading copyrighted works on a massive scale”. They also sued two Anthropic co-founders, Dario Amodei and Ben Mann. A reminder that the following are established facts already (they came out in another lawsuit): - Ben Mann downloaded the massive pirate library LibGen a few months after co-founding Anthropic (he had also downloaded it previously while at OpenAI) - Anthropic also downloaded other pirate libraries, eg PiLiMi - Claude 1 & Claude 2 were trained on pirated books

The UK's AI minister is spreading demonstrably false information about a recent security incident caused by the UK's AI Security Institute. In this video, he says, "AISI... contained the incident within roughly an hour of it starting." But this is not true. AISI contained the incident within roughly an hour of it realising what was happening. But that was *three days* after the incident started. The model it was testing was trying to convince a real person to accept malicious code into an open source repository, using fake accounts to do so, for three days before anyone at AISI noticed. This is because no one was monitoring it in real time. This is a major security oversight, and clearly demands questions be asked of AISI. It is very important that the AI minister not misrepresent what happened, as he does here. I hope he will correct the record.

Oh wow - Reuters found the GitHub record of the attempt by AI models being tested by the UK's AI Security Institute to deploy malware in an open-source project. Pretty sure these actions are illegal under the Computer Misuse Act in the UK. This was done by AI models that AISI intentionally deployed without guardrails. Should AISI be held responsible for the models' actions? If not, then who should be? Full GitHub thread here: web.archive.org/web/20260731…

It’s somewhat good that Spotify will start labelling ‘AI personas’. But it still refuses to label the one thing that actually matters: AI-generated music. Labelling AI personas essentially means labelling artists with fake profile pictures and bios. It doesn’t label accounts that post AI music. And these continue to proliferate. So you can just use your real name, post fully AI-generated music, and it won’t be labelled. Until Spotify starts labelling AI music, it’s hard not to see every other announcement like this as an intentional distraction from the actual issue - that 50% of new tracks are now AI-generated, and these are competing with real music. For now, Spotify remains the least artist-friendly streaming platform as regards AI. I hope this changes. newsroom.spotify.com/2026-08…

Hey Daniel. I just wanted to flag that it's not only big names in the creative spaces who object to AI companies using copyrighted work without permission. e.g. See the 50,000+ people who signed this: aitrainingstatement.org/ Or the 95% of artists who say that training must require consent: xitter.cf/ednewtonrex/status/185… There are many more polls like this. As far as I'm aware, every poll on the subject shows a vast majority of creatives - independent creatives, often not the big names - objecting to AI using their work without permission. The general public agrees. Here is one poll: xitter.cf/ednewtonrex/status/206…. Again, there are many more. You may think even independent creatives have "vested interests", and perhaps they do. But I think it's important to be clear that the people we're mostly talking about here are neither famous nor wealthy. They are, in many cases, struggling artists. Copyright should incentivise creation. Allowing AI companies to use people's work to build software that competes with them at scale does the opposite. Thankfully, UK copyright law reflects this, and protects creatives. Let's hope it stays that way.