Our journal is international, so evidently not every submission is by a native English speaker, heck, I'm not a native speaker either :)

So yes, I'm thinking that if they're upfront about it I'm fine with it.

It's quite a few, more than the occasional one I would get once a year, and I have no idea what has prompted the change. Am I being recommended more in reading lists? Has my writing improved (unlikely), I love untangling new patterns in data, and this has me stumped.

Oh yes, I forgot I had done that! I was in Paris I think.

Some of the events can be fun, but I'm always the resident academic, sort of there for the respectability. Getting paid for a change is nice though 😸

I had stopped going to BILETA style conferences a long time ago, and now I've even missed the last couple of Gikiis, it's tough indeed knowing what to prioritise.

This is more self-aggrandising than my normal outputs, but I’m feeling smug after a good week. Sue me.

The funny thing is that these people have given 0 thought to the legal implications. Training took place in the US, so if they sue in Japan they have 2 options: claim US law applies (currently fair use), or claim Japanese law applies (wide TDM exception). That's why they issued a letter, not sued.

Not enough people have given thought to the complexities of jurisdiction and choice of law in years to come.

But let's say an EU company gets sued in the US for training done in the EU, that is more likely to be contested, and I'd even argue that the AI developer could have a good case that this was legitimate under EU law.

The honest answer is "it depends". There's already been a suit against Chinese model trainers, and I think that this is likely to be decided in favour of the claimant because it will go uncontested. Enforcing that ruling in a Chinese court will be difficult though.

I've told you before, there are three rules of statutory interpretation and I won't be swayed.

These people tend to have a twisted and erroneous idea of the law. There's a big global jurisdiction, all legal systems fit with each other, a fault in one building block brings down the entire legal system, etc.

It's been 3 years, I predicted a 6 year war (10 if there's a SCOTUS case), but by then the technology will be so entrenched that we will get mostly settlements and licensing agreements. It's crazy how many people really believed that copyright would kill genAI.

Pretty frigging huge.

About to start reading it, but browsing the conclusions it appears to be mostly a victory for Stability AI.