I, for one, welcome our new quadrupedal badminton-playing overlords!

All good questions that I'm sure will come up in the lawsuit.

The reason I said the question of specific infringement is "more compelling" is because I think the "theft" question is sort of abstract and might not get very far ultimately. Whereas any IP holder could potentially just go after genAI with a massive "infringement sting". Pay a subscription fee and start prompting in every way you can to get it to generate loads of infringing content, and then sue them for actual infringement.

If the AI companies want to claim that their training is no different than a human reading books and studying art, then treat them like a human that you paid money to create a work of art. If you hired an artist and paid them to make you a bunch of paintings of Disney characters, and they did, would that artist be liable for infringement? If they did it a lot, over a period of time, would that be grounds for a suit larger than simple damages?

I don't know, IANAL.

True, but it gives control of what others do with the material after they read it. For example, if I memorize a copyrighted short story or news article, then I type it into my computer with a couple words changed, post it online, and sell ads against it, that's not "fair use" is it?

This is basically what newspapers and authors have proven genAI can do - using prompts to basically recreate book passages and article almost verbatim.

It's the same thing - monetization of other peoples' works without permission.

Exactly my point. The business model isn't sustainable in its current form in a legal way - hence them being sued over and over. It's about time.

If you use an SaaS tool that you pay for and then simply ask it to generate a spline that "looks like a nike swoosh" then it might be infringing, because the company would have to have programmed the tool to a) know what "nike swoosh" means, and b) to recreate something resembling it. If the tool had never seen the logo and hadn't been trained to associate it with a brand, then it wouldn't be able to generate what you asked for.

There's a huge difference between a blank canvas where you as an artist have to move a pointer around in a path that replicates an IP, and a tool where you simply type a few descriptive words and it creates the image for you.

If you want to take the analogy to a ludicrous degree, there are some "tools" that are deemed illegal or inappropriate for the public market because despite having "legitimate" uses, they make criminal activity too easy. Lockpicks and signal jammers are not allowed in many places, automatic weapons are banned almost everywhere, and color printers that can perfectly duplicate money, for example.

I'm not saying Midjourney is a "dangerous weapon" but in their current state it's a tool that can very easily facilitate infringing activities on a mass scale, so there's precedent to look at these issues from a legal standpoint.

You may be right, but it doesn't really matter at this stage of generative technology. There's apparently no way for these companies to put up guardrails that are both strict enough and competent enough to prevent infringement without also making them useless. They'd need to literally hire people to review every generated image to see if it meets the required test of "does this look enough like Moana (et al) to be infringing".

The only other option would be to completely retrain new models using only imagery in the public domain, or licensed specifically for this use.

The US copyright office's recent report concludes that while some generative AI probably does constitute a "transformative" use, the mass scraping of all data for commercial use probably does not qualify as fair use.

"The extent to which they are fair, however, will depend on what works were used, from what source, for what purpose, and with what controls on the outputs — all of which can affect the market," the report states.

"Making commercial use of vast troves of copyrighted works to produce expressive content that competes with them in existing markets, especially where this is accomplished through illegal access, goes beyond established fair use boundaries."

The GenAI companies love to compare their "training" models to a regular art student "training" their talent by looking at existing art.

The difference is that a human artist can't then turn around and take commissions to sell paintings of other people's IP, right? You can't just learn to draw Moana and then sell Moana TShits online - that's infringement. But Midjourney will take your money for access to their models and serve you up infringing images.

I don't think the "theft of IP" angle is quite as compelling as the "selling infringing material" side is. But the end result is the same.

Sorry to be blunt, but this is just laziness. I hit the political fatigue wall a few months ago, so I made a new feed that only includes accounts that are focused on specific non-political topics - dogs, games, etc. Took me all of 5 minutes and that's my default feed now. The "following" feed is still there when I want to dip my toes into the wider garbage-sphere, and liberal use of easy to find block lists keeps the trolls away. The difference is that I don't spend all day scrolling because I actually run out of stuff to look at on Blue Sky and shut the browser tab when I'm done. It's delightful!

I don't know if their business model is sustainable, but it would be a real shame to lose it, because personally this bubble is exactly what I'm looking for from social media right now. I have zero regerts over not hearing the toxic bullshit and slop that's taken over the other platforms because I'm not a drama-addicted shell of a human that needs to "engage" with random strangers that have nothing but stupidity to contribute.

I can have meaningful conversations about topics with people of different viewpoints by engaging with my extended friend group and a curated list of news sites and subreddits.

Is this actually a ploy for Texas to make Apple collect all the data so their politicians can later subpoena that info for their own nonsense?

I do wonder how loose the requirements will be for proving "parentage" and "approval".

Will a credit card suffice? If you login to your kid's iPhone (or your little brother's) and enter your credit card info and give them the PIN, does that count as tacit approval? Or will Apple have to call the parents for voice approval on every purchase?

I tried ARC and thought it was pretty cool - but boy I'm just too busy to learn a new browser. I probably also am not the target market, since I mostly use a browser to, um, browse the web, and dedicated apps for 75% of my productivity work.

Sometimes "good enough" is too hard to replace.

Wouldn't these companies say that it's all "opt-out by default" because by "default" you aren't doing business with them?

By doing any level of business with a company you are, in effect, opting-in. Just like when you shake a stranger's hand you're opting in to viral exposure, and if you walk down the sidewalk you're voluntarily exposing yourself to pollution and dirt on your shoe.

When you deal with a company you have to either do a bunch of research to understand what you're opting in to, or you just have to assume the absolute worst. Neither is something the average consumer wants to have to do, which makes sense.

The way the current administration is completely dismantling consumer protections in the US I don't have much hope for this movement here, but I keep dreaming anyway.

I was telling people this 30 years ago when I was shilling Cutco knives - it was one of their selling points.

Giving someone a "Below Expectations" rating because upper management wants to hit numbers is such a crappy mind game. If everyone on your team is performing to the expectations you set out for them, but you need to reduce headcount, just rate them as they deserve and lay them off based on some other objective criteria, like seniority or cologne preference.

This feels like a corporate CYA move to avoid wrongful termination suits by lying about staff performance. I wonder if leaks like this could be used for a lawsuit instead...

Maybe people who launch and close and relaunch office apps all day will benefit from this. But most people I know don't work like that. You start the day by bringing up all your main apps and leave them running all day.

But as usual, MS loads more garbage on your system without asking and enables it by default, leaving it to users to figure out what changed and how to disable it.