Azuaron

Anthropic--the AI company that's actually doing "well" out of the pack--lost $42 billion dollars last year. I'm just going to say it: these aren't companies. Companies don't lose $42 billion dollars in a year without going bankrupt. I don't know what the "AI companies" are, but I won't even concede them the name "company" anymore.

@akareilly@hachyderm.io Linux is very much a "socialize the costs, privatize the profits" kind of tech. Edit: Folks, if you're going to "well-actually" me about this, first of all I guarantee you haven't actually thought about this as much as you need in order to do that. Second of all, this was a quip not a detailed technical analysis. I'm not going to write you a detailed technical analysis unless you pay me, and you can't afford the nuisance charges I would require to do that. No, you can't trick me by saying something stupid and wrong in an effort to get me to correct you, so just keep your stupid and wrong nonsense to yourself.

@risibledog@infosec.exchange Tangential thing I learned recently: "sabot" is a French word for a wooden shoe. "Sabotage" is when you throw that wooden shoe into the mechanical loom, destroying it, because the bosses can fuck all the way off.

@danielcornell@mastodon.social Please be Sam Altman. Please be Sam Altman. Please be Sam Altman. Please be Sam Altman.

@mttaggart@infosec.exchange There are a bunch of separate things at play, some of which are pretty cut and dry, some are less so. tldr; I think Microsoft/OpenAI are facing several violations: Unauthorized access of the NYT servers (computer fraud)Illegal acquisition of NYT articles (copyright)Potentially infringing LLM output (copyright)Violation of the Hot News Doctrine (quasi-property rights) The first part is "how did they acquire the content." Depending on the method of their "hack", they might be guilty of not a copyright claim, but the Computer Fraud and Abuse Act for unauthorized access of a computer system. Of course, maybe they get hit for both some kind of hacking and also the copyright. Anthropic got in trouble for illegal acquisition, but their acquisition was via torrenting, which is a legally well-defined copyright violation. It's unclear to me how "hacked a computer system" would make the illegal copyright acquisition go away, so it seems like Microsoft/OpenAI should get hit for the copyright issue here just like Anthropic did. And, of course, the judge in the Facebook case went full insane and said torrenting was legal for Facebook, so anything's possible, I guess. After the acquisition, we've got the training. "Training" is generally considered to be Fair Use, in and of itself, since "training" really just means "collecting mathematical statistics on word sequences", and we have decades of legal precedent saying that you can collect mathematical facts regarding copyrighted material and it's Fair Use. Finally, there's the question of distribution. Generally speaking, the question of LLM distribution usually doesn't reach the Fair Use question, because it doesn't pass the "substantial similarity" test: something is only a copyright violation if it is "substantially similar" to the copyrighted work, and LLM output is usually not similar enough to any individual copyrighted work for LLM output to be a copyright violation. However, that's generally, so let's talk about a case out of Germany. There was a music AI company that musicians claimed was outputting substantially similar work, and the AI company claimed that it was because the prompters were "steering" the AI to generate copyrighted works, so the prompters should be liable. The judge's ruling came down and it did ultimate hinge on that steering question: if the prompters could be said to be steering the AI to reproduce copyrighted work, then the prompters would be responsible for the violation, but if the prompters were asking for something generic and the AI was providing substantially similar music as output, then the AI company would be liable (in this case, the judge ruled that the AI company was liable). It should be further noted: the judge in the Anthropic case specifically said if Claude was producing substantially similar output, then Anthropic would be liable for that. So it seems pretty likely that if the LLM is reproducing copyrighted work, and the prompters aren't "steering" it, the AI company is liable for copyright infringement. Apart from the steering, though, why was the AI outputting substantially similar music? Well, in the case of word-based LLMs, they're drawing on trillions and trillions of inputted documents. There's so much less music. So, if you prompt for a less-than-popular genre and a couple of "with this kind of beat" or whatever, the AI doesn't have a lot of music to draw on for its response, so often ends up outputting music that's substantially similar to copyrighted music. In the NYT case, I don't think they're going to get away with the "prompter is steering it"; people are going to put in prompts like "what's going on with ____" or "tell me the latest news", and they're going to get information straight out of a small number of publications, of which NYT is a significant part. Since all the information is coming out of a small number of publications, the AI companies are going to have a similar problem as the music AI: there's only so much mixing the LLM can do with just a few sources, and the output's likely to be substantially similar to at least one of their sources. But even if the LLM output is not substantially similar, there is still the Hot News Doctrine. Essentially, during WWI the AP was gathering news from the war and publishing it on the east coast, and the International News Service would take those articles, telegraph them to the west coast, and beat the AP to publication over there. The courts ruled (and have affirmed several times, including recently) that even though facts were not copyrightable, the AP had "quasi-property rights" over the facts that they had expended a lot of time and energy finding and determining, and the INS' free-riding on the AP's hard work constituted unfair competition. There is a five element test for determining a violation of the Hot News Doctrine: 1) A plaintiff generates or gathers information at a cost; 2) The information is time-sensitive; 3) A defendant's use of the information constitutes free riding on the plaintiff's efforts; 4) The defendant is in direct competition with a product or service offered by the plaintiffs; 5) The ability of other parties to free-ride on the efforts of the plaintiff or others would so reduce the incentive to produce the product or service that its existence or quality would be substantially threatened. It seems incredibly likely to me that the AI companies are violating the Hot News Doctrine. They're literally admitting point 5, which seems to be the point of most contention in previous cases. Indeed, last year, in this very lawsuit, the judge throw out a claim by NYT regarding the Hot News Doctrine because OpenAI was only training on the articles, not, in a timely way, distributing the reporting. I don't think that's still the case; if ChatGPT is outputting timely news that it didn't get a license for, the judge explicitly said that while training was not Hot News freeriding, scraping and then turning around and outputting LLM news of the scraped information could be. I am, also, not a lawyer, I just have a hobbyist's interest in this kind of thing, so grain of salt and all that.

@tante@tldr.nettime.org @frameworkcomputer@fosstodon.org I need the Apu jumping in front of a bullet meme for Framework and Framework's weird nerds inventing benefits of the doubt for a guy that was so racist, he thought London needed to go back to being just ethnically Briton. Not just white, but Briton.

@futurebird@sauropods.win Definitely makes me think of alien face huggers, but apparently it's some people's name for this emoji: 🤗

@redsad@ohai.social Me: *braids a whip* Them: What's that for? Me: Just heading to church. An author's giving the sermon, and he's definitely going to be selling copies of his book in the narthex.

@FeloniousPunk@beige.party He wants to charge people to get his messages early, and I'm wondering if I can get paid to get them late--or, better yet, never.

@jplebreton@mastodon.social Tired: The "web". Delicate. Must be rebuilt every day. Single-owner. Wired: The mycelium. Network of many. Hidden. Lives on the decay of the living. "Shit-post" has an additional meaning. "You cannot kill me in a way that matters."

@henryk @codinghorror I used to think the "just copying from Stack Overflow" jokes developers made were, you know, jokes. Then I watched all my coworkers embrace LLMs, and I was forced to conclude that they weren't joking. So, depressing agreement.

@inthehands@hachyderm.io I believe the response to their long reply should be a simple, "I ain't gonna read all that, taker. Here you are trying to take even my time."

@Beggarmidas@mastodon.social @EdCates@mastodon.social @selzero@syzito.xyz "Yelling at me isn't going to change anything," says man yelling at people like it's going to change something. "You have an enemy that is taking full advantage of your sides tendency to fragment," says man trying to fragment the left off his side. "Accelerating it. Feeding it oxygen. Actively provoking you to make irrational decisions..." says man who thinks floating the same kind of candidates that have spent my entire adult life getting the shit kicked out of them in elections is going to change things. Doing the same thing and expecting different results, as they say. "... and make war among your fellows instead of alliances & coalition build," says man actively trying to destroy any alliance with people to his left. "Further enhanced by the Russians actively efforts to trigger all grievance politics into full bloom." Honey, you don't need the Russians to fuck things up. Clearly. You're doing just fine all on your own. Have you ever introspected in, like, your entire life? JFC. :neofox_laugh_tears: Get your fucking priorities straightened out, King Hypocrite.

@cR0w@infosec.exchange I mean, they say this shit every few months. I'm surprised they can still get press for it.

@cR0w@infosec.exchange I know they're doing it for the investors. I just don't know why the investors keep falling for it.

@mttaggart@infosec.exchange I've started thinking about this kind of thing like cabinetry. There are beautifully crafted cabinets made by experts with a lifetime of experience... and there's Ikea. Businesses have always wanted the Ikea version of code. Edit: lol, the Ikea defenders have arrived. Don't bother. If you think Ikea is selling anything of quality, that's lack of experience on your part and nothing to do with me.

@qurlyjoe Thing one: Why is the name censored? They're starting a business, gotta get that name out there! Thing two: But actually, I can read that name under the censor bar. That is not a particularly effective censor bar.

@ana Time to make the distributed system more distributed!

@alan@mindly.social @alexmdunne@mastodon.social @carnage4life@mas.to Not that I like AI, but I literally had a human coworker that drove me crazy with shit like this.

@timberwraith@mastodon.social There's a lot of bullshit around what the Democratic party is/has always been doing, but I've gotta point out that they want "manly man" candidates so they got a guy named "Poindexter", which is literally the classic name chuds pull out to bully wimpy nerds.

@matthew@opinuendo.com "They should be held accountable," said a police captain, as if he wasn't the problem.

@IzzyOnDroid@floss.social Adults have learned critical thinking? Citation needed.

@cR0w@infosec.exchange Holy shit. Last time the Senate passed it and the House didn't vote. Are we just doing the reverse now?

@futurebird I found their Amazon page and gave them a glowing (like nuclear waste) review, and reported their product page. Will Amazon do anything? Probably not. But hopefully it makes them do some damage control, at the very least.

@gloriouscow@oldbytes.space @petrillic@hachyderm.io My friend, that's not an argument against what they said.

@cR0w Blue Team practically working for Red Team.

@markmccaughrean I love the skim of atmosphere, particularly where it's not so bright.

@ohmu@social.seattle.wa.us @mekkaokereke@hachyderm.io I would say it's not as egregious as The Giving Tree, but it's not not a Giving Tree story.

@oscarjiminy@aus.social @mekkaokereke@hachyderm.io I haven't heard anything about Silverstein being a sex pest. I have heard he was a pretty terrible husband and father, but in a non-criminal way. The Giving Tree itself, though, is a story about how a boy (then man) abuses a tree until he finally takes the entire tree, utterly destroying it. Many people see that as an uplifting and inspiring story about the sacrifices of motherhood/parenthood; and many other people see the narrative that it's supposed to be an "uplifting and inspiring story about motherhood" as toxic patriarchy rearing its ugly head and demanding literally everything from women. And many people see Silverstein as a bit of a troll who did not actually mean for the story to be uplifting and inspiring, but more of a cautionary tail against being a doormat for your children.

@ethanschoonover Dealer's choice.

@goatrodeo@mstdn.social Please tell me you didn't just start today...

@theleftistlawyer@mastodon.social I'm still wondering how Native American reservations are the appropriate thing to do, instead of just full-on Native American states. Isn't like half of Oklahoma the Cherokee Nation now? That should just be a state. Give them two Senators, House reps, the whole nine yards. Repeat.

@allpoints@mstdn.social @quixoticgeek@social.v.st You know in 80s movies of the 50s when the macho guys would drag race toward a cliff, and the goal is to get to the cliff, first, and get closer to the edge of the cliff than anyone else? Sometimes by stopping the car, sometimes by bailing out of the car. That's what they're doing. But with the economy.

@mekkaokereke@hachyderm.io I've been in too many conversations about The Giving Tree, and now stories like this have been preemptively poisoned. It certainly does not help that this one features a tree.

@gmh @cstross Just so we're clear about what's happening, men are already required to register for the draft within 30 days of turning 18. Any who don't can be hit with fines up to $250k and 5 years in prison, as well as a prohibition against receiving lots of federal funds (such as student aid). Automatic registration will at least close the danger of missing that window accidentally. "You have to do this thing by this day, we won't necessarily tell or remind you about it, but we'll punish you if you miss the date," are some of my least favorite laws, and it's not like failing to register for the draft would actually let you dodge a draft. All that being said, the draft is slavery and if a draft is instituted literally every draftee should say so and refuse all orders while suing the government.

@ZachWeinersmith I mean... I can think of a few necessary actions that I would take if I actually thought super-human AI was coming next year...

@mainframed767@infosec.exchange Ah, well, not entirely. What they're really for is providing "auditing" services that get them everyone's data in an industry, and then turn around and offer consulting services to the companies in that same industry using all the data they got. This allows industries to, for example, price fix without illegally colluding as a cartel. "We didn't collaborate with our competitors; we just did what the consultants told us to! And the consultants told all our competitors to do the same thing!" You might remember the bread price fixing scandal in Canada a few years back where I believe it was McKinsey was the go-between. AI is a threat to this price-fixing as well: all the companies are feeding all their data into the LLMs, so they can get similar responses from the LLMs. Anyway, even if all BCG or McKinsey did was "present the CEO ideas as the path forward or justify layoffs", AI is still a threat to these consulting companies because the CEO can also get that nonsense from the AI.

@loriemerson@post.lurk.org ...I've never before considered converting to Catholicism, but now I'm thinking about it.

@scottwilson@infosec.exchange Do you have a Ubiquiti replacement recommendation?

@anne_twain@theblower.au @broadwaybabyto@zeroes.ca This is the guy that promised to solve world hunger if the UN World Food Programme could give him a plan to do it. They gave him a plan to avert famine for 42 MILLION people, and he ghosted them.

@jztusk@mastodon.social I hate AI more than just about anyone... but are you saying that I need to ask for consent to read your public website? Should I have DMed you first before reading this toot? What point are you trying to make here?

@cR0w@infosec.exchange Linus once told a room full of Google engineers that they were idiots because he disagreed with them about how they setup their code repository. He's never had a brain-to-mouth filter.

@mondoweiss@social.mondoweiss.net Let's not forget about Trump's own war crimes that he wants to be shielded from.

@mo@mastodon.ml First of all, that's not what they said. They said "I have downloaded" data that "you were probably expecting to be yours", as if the speaker of the poem hacked into a secure system and stole data. Second of all, "copying" is how the internet and computers work, you can't get around that. Thirdly, now you're making a copyright argument--which, again, the poem is not--and a copyright argument is never complete without Fair Use, and every Fair Use analysis of training AI models concludes--often begrudgingly--that training AI models is Fair Use, including the courts that have ruled on it. I also can't, personally, come up with a way to revoke Fair Use for AI that wouldn't revoke Fair Use for a ton of excellent Fair Uses, because it turns out that "AI" is a marketing term, not a technical one, and at the end of the day there's a lot of excellent and fair uses for "scraping data off the internet" and "programmatic remixing of data". A while back, there was someone on Twitter that I think even built a non-profit out of screenshotting ads on nazi sites like The Daily Stormer, and contacting the corporations in the ads saying, "Hey, do you really want your name and logo right next to 'Women shouldn't be able to vote' and 'Cops should kill more black people'"? Most corporations didn't, obviously, and this effort cost The Daily Stormer and similar sites lots of money. Millions of dollars. Obviously, The Daily Stormer hated this, and would never provide consent for it. I believe this effort was primarily a manual one; people would load up the target sites, take screenshots, and share the screenshots around. Let's say I wrote an application to automate this. With an embedded browser, it loads up the sites, takes screenshots, and then does it again and again, varying browser agent, referrer, and IP address with a VPN so it doesn't get blocked and it gets different ads. Then, I can analyze the images to eliminate duplicates, automate determining the brand in the ads, and create mosaics of brands next to harmful content, and then I shared those mosaics in the same way that the screenshots were shared on Twitter to shame ad companies off the sites. What I just described is a more direct copyright violation than what LLMs do, with less Fair Use protection because the end result contains directly copied portions of the copyrighted work. If you think I shouldn't be allowed to do that, then we fundamentally disagree about copyright law in general, and there's probably no point in this conversation; you literally want to get rid of Fair Use. If you think I should be able to scrape and collage The Daily Stormer as part of a hostile campaign against them, how do you craft a law that outlaws "AI" "training", but still lets me do that? Because, again, "AI" is a marketing term, not a technical one, and the law has to be technical. Finally, copyright law is primarily law wielded by large corporations against individuals. I am a programmer and an author; I spend all day every day creating copyrighted work. I have never been helped by copyright law, only harmed by it. I don't know about you, but I don't want to hand Disney--or my employer--yet another club to bash people with.

@ErikJonker@mastodon.social To be absolutely clear, the US military has already obeyed orders to hit civilian targets in Iran. That's already happened. A lot.