A damning new study could put AI companies on the defensive.

In it, Stanford and Yale researchers found compelling evidence that AI models are actually copying all that data,
not “learning” from it.

Specifically, four prominent LLMs
— OpenAI’s GPT-4.1, Google’s Gemini 2.5 Pro, xAI’s Grok 3, and Anthropic’s Claude 3.7 Sonnet
— happily reproduced lengthy excerpts from popular
— and protected
— works, with a stunning degree of accuracy.

They found that Claude outputted “entire books near-verbatim” with an accuracy rate of 95.8 percent.

Gemini reproduced the novel “Harry Potter and the Sorcerer’s Stone” with an accuracy of 76.8 percent,

while Claude reproduced George Orwell’s “1984” with a higher than 94 percent accuracy compared to the original
— and still copyrighted
— reference material.

“While many believe that LLMs do not memorize much of their training data, recent work shows that substantial amounts of copyrighted text can be extracted from open-weight models,”
the researchers wrote.

Some of these reproductions required the researchers to jailbreak the models with a technique called "Best-of-N",
which essentially bombards the AI with different iterations of the same prompt.

(Those kinds of workarounds have already been used by OpenAI to defend itself in a lawsuit filed by the New York Times,
with its lawyers arguing that “normal people do not use OpenAI’s products in this way.”)

The implications of the latest findings could be substantial
as copyright lawsuits play out in courts across the country.

As The Atlantic‘s Alex Reisner points out,
the results further undermine the AI industry’s argument that LLMs “learn” from these texts
-- instead of storing information and recalling it later.

It’s evidence that “may be a massive legal liability for AI companies”
and “potentially cost the industry billions of dollars in copyright-infringement judgments

futurism.com/artificial-intell…

This entry was edited (20 minutes ago)

Call it hashtag stacking or just spamming, but addding 20 hashtags to your post is super annoying to screenreader users, since the screenreader reads out every single hashtag of yours. so, please, add your hashtags to end of your post, even preferably after an URL. And who ever of the app developers out there creates a feature which detects this pattern in posts and allows me to disable it, has my vote.

I lost a friend over my article about the murder of Renee Nicole Good.

He was angry I called out the underlying misogyny.

He was mad I didn’t give ICE the benefit of the doubt.

He made it personal.

He told me my “stupid clown hair” is why people don’t take me seriously

Everything he said to me was dripping with disdain and misogyny.

There was no debate.

There was no respectful disagreement.

In under five minutes he had attacked my intelligence, my politics, my appearance and my writing.

This is not normal behaviour.

He literally proved my point.

Had I taken the bait I’m convinced that I would have been called a “f*cking b*tch” by the end of the conversation.

This is what MAGA is doing to men.

It’s emboldening them.

It’s validating their bigotry.

I assume this person was always hateful and misogynistic on some level … but now he no longer feels he has to hide it.

He literally said as much.

He said “I never told you in the past because I wasn’t allowed. Now I am.”

Why is he allowed to tell me now??

Why does he feel justified in lashing out with rage and insults over something that has nothing to do with him?

It’s because of what happened to Renee Nicole Good.

We all watched a woman get shot in broad daylight and there’s been zero consequences

We’ve watched rapists, pedophiles and abusers go free for years.

Every woman who’s ever experienced abuse knew exactly what Jonathan Ross meant when he said “f*cking b*tch”

He didn’t fear for his life.

He was enraged she didn’t fear for hers.

My friend was enraged I dare speak out for a “lesbian leftist”.

He was enraged I didn’t remain silent, complicit and demure.

He was apparently also enraged I didn’t realize he hated my hair and change it to please him.

Make no mistake, there is a war against women going on.

This regime wants to make us “less than”.

They want us afraid.

They want us in the home serving our husbands and having children.

They want us seen and not heard.

They don’t want us resisting.

Which is why we must keep speaking out.

We must refuse to go back.

My life is better off without this person in it, and thankfully I can walk away.

Not everyone can.

Speak up if you see someone being hurt.

Protect your friends and neighbours.

Challenge misogynistic talking points.

Demand justice for Renee Nicole Good.

Don’t cower. Don’t be afraid. Don’t give up.

#uspol #fascism #reneenicolegood #abolishice #immigration #minnesota

The measles outbreak in South Carolina is the largest in the country and has spread to three other states.

It’s growing at an alarming speed.

Over 500 people in quarantine and 200 are actively infected.

This was completely preventable.

These people didn’t need to get sick.

Measles can kill. It can disable. It can leave you deaf and blind.

It’s a vaccine preventable illness.

Shame on everyone creating distrust in the vaccine and refusing to do their part for public health.

nbcnews.com/health/health-news…

#uspol #measles #publichealth #disability #ableism

haha. I have made blindsoft image describer-js (BSID-JS) open source. #opensource #cloudflare #workers #js #javascript github.com/averlice/bsid-js if you want to host or contribute at all, feel free. licensed under the gpl-v3 public distribution license.

Started watching BSG again today as I wanted to see if it looked as awesome on a 77" OLED TV as i thought it would (it does) and, oddly, it has over the years become something of a comfort show.

It continues to impress me how well Ronald Moore and Co. dealt with issues touched upon 20 full years ago that continue to be relevant today.

tvscholar.com/p/20-years-later…

Blind people now have AI image description and these days a Blind person will send another Blind person just an AI description of a selfie, with no source photo. In the complex AI discourse i would just like to say that AI image description has changed the game for Blind folks. It's a hell of a ride at age 43 to read my first picture book and first graphic novel. AI is complicated but textualizing images is brilliant