This is getting stupid. It's like saying "we checked your travel time and we suspect you didn't walk under your own power for the entire journey". Who gives a fuck?

By Anthropic's description, if I write thousands of words of my own work and ask Claude to check for any spelling mistakes, and Claude says "one typo, you wrote CIa instead of CIA, fixed" what comes back has the watermark.

NOBODY KNOWS what Copilot apps they're using. Microsoft Copilot? Microsoft 365 Copilot? Copilot chat (available to Microsoft 365 subscribers)? Microsoft Copilot Studio? Github Copilot? Microsoft Copilot for iPhone?

beat me to it.

How did Microsoft blow this? It had everything it needed to be number one in everything to do with AI, and now its only users are people who don't have a choice.

Freenet is also based on middle-out compression.

No. Trying to push 30TB into Freenet will be interpreted as an attack on the network by other peers and they'll disconnect from your peer.

Freenet is better viewed as a communication medium than a storage medium. Data is prioritized according to demand, somewhat like an LRU cache. You can't upload 30TB and expect the network to preserve it for you.

On liability, running a Freenet peer is more like running network infrastructure that automatically routes and caches other people's traffic than intentionally publishing it. US law explicitly recognizes this distinction in the DMCA's provisions for transitory network communications and system caching. We discuss this in more detail in the FAQ.

You obviously haven't tried Freenet. You can join our room on River and you'll see no spam, no fraud, no hate speech, none of that.

Decentralized doesn't mean unmoderated. Systems built on Freenet can be moderated however their creators and communities want. River is moderated.

The difference is that nobody running one service gets to decide what everyone else is allowed to run.

Now that all this information is valuable (to AI bot scrapers), it's all being gatekept from humans who refuse to log in and participate in generating more valuable content for AI bot scrapers.

It's basically Farmville (but for scrapeable content), but restricted to the existing domesticated population of users. If you refuse to have a collar put on you, you no longer are allowed to access an echo copy of information that used to be on the open web before SEO garbage displaced it all.

Surfing with javascript turned off to help defend against drive-by 0-days? Half a dozen anti-bot systems either block or temp ban you. Use the wrong IP address, and reddit prevents access.

You could say this is a natural evolution of order in a capitalist society, as explained by Cory Doctorow's views on enshittification. However, I would argue this goes back even further to RMS' views on open vs closed software. Essentially, when no money is at stake, people are willing to share. When money is at stake, people start putting the things people were sharing behind closed doors and charging for entry.

If you extend the open source metaphor, being able to fork your posts theoretically allows individuals (and by extension a community) to transcend greed and censorship. Consider how much historical content (both virtual and IRL) has become cancellable in the last 4 presidential administrations, depending on which side of the current wave of bandwagoning you stand on. Even before that, how much content was rejected or "rewritten" on spaces like Wikipedia to satisfy the personal views of editors who staked out and camped specific areas on Wikipedia. Mind you, this was before Wikipedia became a force multiplier for influencing, which led to attacks by astroturfers and state actors.

Which then leads me to the question... what forkable, censorship resistant alternatives to reddit exist? I don't want to have a brand just to have my own legion of supporters to defend and reinstate me, or rely on glacial bureaucracy mandated by law if I get deplatformed just for having a conversation just because some AI bot had a bad day.

While what you say may be true for state and local law enforcement, it is not true for federal law enforcement.

https://www.mololamken.com/kno...

"Under Section 1001 of title 18 of the United States Code, it is a federal crime to knowingly and willfully make a materially false, fictitious, or fraudulent statement in any matter within the jurisdiction of the executive, legislative, or judicial branch of the United States."

Actual text of the code:

https://uscode.house.gov/view....

"Â1001. Statements or entries generally

(a) Except as otherwise provided in this section, whoever, in any matter within the jurisdiction of the executive, legislative, or judicial branch of the Government of the United States, knowingly and willfully-

(1) falsifies, conceals, or covers up by any trick, scheme, or device a material fact;

(2) makes any materially false, fictitious, or fraudulent statement or representation; or

(3) makes or uses any false writing or document knowing the same to contain any materially false, fictitious, or fraudulent statement or entry;

shall be fined under this title, imprisoned not more than 5 years or, if the offense involves international or domestic terrorism (as defined in section 2331), imprisoned not more than 8 years, or both. If the matter relates to an offense under chapter 109A, 109B, 110, or 117, or section 1591, then the term of imprisonment imposed under this section shall be not more than 8 years."

This is the statute under which Martha Stewart was sentenced to federal prison.

https://www.nyccriminalattorne...

"The interview isn't about solving the crime they're investigating. It's about creating a new one. When federal agents can't prove the underlying offense, they charge you with lying about it. Martha Stewart wasn't convicted of insider trading - she was convicted of lying to investigators about insider trading she was never found guilty of. The crime she went to prison for was created in the interview room. This is the federal playbook: investigate one thing, charge another. The conversation itself becomes the crime.

18 USC 1001 makes it a federal felony to make false statements to government agents. Five years in federal prison. $250,000 in fines. No oath required. You don't have to be in a formal interview room. You don't have to sign anything. A casual conversation on your front porch counts. Any federal matter, any federal agent, anywhere. And here's what nobody tells you: the agents asking questions usually already know the answers. They're not asking to learn what happened. They're testing whether you'll tell the truth."

I am also not a lawyer, so your mileage may vary.

Thanks for the heads up.

Have you tried setting up something like a whitelist (see below)? I feel like that would make my life a lot better - just having a process that filters out the obvious SEO sites so that I don't have to do it.

https://github.com/searxng/sea...

Two words:

Parallel construction.

https://en.wikipedia.org/wiki/...

"An example from one official about how parallel construction tips work is being told by Special Operations Division that: "Be at a certain truck stop at a certain time and look for a certain vehicle." DEA would alert state troopers and they may wait for that certain vehicle and use drug searching dogs to identify illegal drug-related activity, giving the appearance the search was conducted randomly.[4] "

This example also would be a form of evidence laundering. By the time the case is presented to courts, the fact that this all resulted from a warrantless search would be nowhere to be found.

https://techcrunch.com/2026/07...

"The payout will deliver $3,000 per work across an estimated 500,000 works, shared among the authors and publishers who hold rights to them. While the settlement is believed to be the largest in the history of U.S. copyright law, many authors and creators still donâ(TM)t view it as a win.

Thatâ(TM)s because of how the legal question was resolved. Alsup sided with Anthropic on the core issue. He ruled that training an AI model on copyrighted text counts as fair use â" a decision widely seen as a turning point for the AI industry. But the ruling didnâ(TM)t excuse how Anthropic obtained the books in the first place. Anthropic had built its training library from two sources: books it purchased and scanned (fine), and books it downloaded from pirate sites like Library Genesis and Pirate Library Mirror. Alsup found the second method illegal on its own terms and said that piracy question could go to trial; Anthropic agreed to a settlement soon after to avoid a trial and whatever damages a jury might have awarded."

So... it's legal to scan books you own and then use them to train LLMs, but it's not legal to use scans that someone else made (I'll assume in this case, they didn't own the books in question.) Hence... the perverse incentive to buy and re-scan books that might already have been scanned... and the cheapest way of doing it is to chop the spine off.

"Internal Anthropic documents about its plan to scan millions of books, revealed in the copyright lawsuit, donâ(TM)t make clear why the company wanted to destroy the books in the process. A deposition of Tom Harvey, who Anthropic hired to lead the project and who previously helped create Google Books, shows that one company Anthropic contracted to scan the books was Datamation, which offers both âoehigh volume destructive and non-destructive book scanningâ services. In a destructive book scanning process, the spine of the book is cut so the pages can be fed into a scanning machine, which is faster and cheaper than non-destructive book scanning.

Regardless of its original intentions, the federal judge in the copyright lawsuit from authors against Anthropic, William Alsup, found that Anthropicâ(TM)s creation of digital copies of the books was legal specifically because the books were destroyed.

âoeHere, every purchased print copy was copied in order to save storage space and to enable searchability as a digital copy,â Alsup wrote in his ruling. âoeThe print original was destroyed. One replaced the other. And, there is no evidence that the new, digital copy was shown, shared, or sold outside the company.â "

Kind of fucked up that the scan can't be shared (or donated). I imagine in most cases, good copies of these books no longer exist in libraries or in the Library of Congress. With current law, all books that are covered under copyright will eventually fall into the public domain, but this is meaningless unless copies exist for people to redistribute once that limit is reached. Essentially companies are exploiting the monopoly benefit extended through copyright without allowing society to benefit from the material falling into the public domain, which is the implicit contract to using state power to enforce copyright.

Ironically, destroying physical copies in order to comply with the 1:1 rule makes the remaining copies that much more valuable.

My experience thus far with the various LLM chatbots is that they've all been trained with the default idea that users want to one-shot solutions.

A human in a similar situation normally would ask questions to clarify the problem, and attempt to ascertain the level of domain knowledge so that they can intelligently communicate.

Part of the song and dance in forming a prompt is to pre-load the model with all those items, so that you don't burn an obscene amount of tokens for it to finally get to the point where it can be useful. An open ended query can be useful for discovery but only if you're willing to do the legwork to cross-reference against other models and reference material.

To put it another way... most humans are lazy and just want to put in the minimum of work, and are willing to accept a 60% acceptable answer from a bot (basically using it as a search engine replacement). Add the fact that most humans are not domain experts, nor are they able or willing to educate themselves to a minimum standard, they have no way of judging whether the bot answer is even at the 60% standard - instead they do what most people do... if it sounds good, and the bot claims to be an "expert", then they go with that.

I anticipate that models specifically curated for specific domains will become more and more useful. I've seen models plugged into residential and commercial electrical code, for example, to act as an always available code compliance helper for a specific municipality. And, as you point out, as long as you can look up the relevant section of code to verify that the LLM supplied advice is correct, and do the load calculations to make sure the math is correct, you can validate the LLM's answer.

The demand is real... whether they can scale the workload down to actually make money remains to be seen.

Long term there are major benefits to society if they can drive down the cost of energy, cooling, and computational infrastructure to support democratized access to LLMs and diffusion models without bankrupting providers.

In the meantime, I would just settle for a decent search engine that didn't require me to go to page 3 of search results (and forcing me to verify that I'm a human for daring to click to page 2 of the search) to find a relevant link that wasn't SEO spam. You'd think that a university agricultural reference for the public would be in the top rankings when searching for a specific variety of vegetable, but apparently today, that's no longer the case.