Anthropic News & Discussions

[weatheriscool]

I wish they'd just keep it where it is. Limit the other lower models if need be.

[firestar464]

[firestar464]

Big if true, though it is important to note that this particular problem has a history of epic fails.

[firestar464]

yeah, it's now widely accepted

https://removepaywalls.com/https://www. ... -surprise/

[firestar464]

As reposted by Jimmy.

[firestar464]

[weatheriscool]

[wjfox]

Introducing Claude Opus 5

24 Jul 2026

Claude Opus 5 is available today. It’s a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price.

On coding and knowledge work evaluations like Frontier-Bench and GDPval-AA, Opus 5 is the new state-of-the-art, though it remains behind Mythos 5 on cybersecurity tasks.

Opus 5 is designed to be used every day: it works more efficiently than other models. It’s the new default model on Claude Max, and the strongest model on Claude Pro.

https://www.anthropic.com/news/claude-opus-5

[weatheriscool]

[firestar464]

Snopes reviewed internal documents from Anthropic that show they have bought millions of books to "destructively scan" them to train their AI model.

https://www.snopes.com/fact-check/ai-co ... are-books/

Ok as long as the book is not extremely rare or national heritage or whatever. No evidence that they're destroying actual rare books though, so the outrage is premature imho

[Yuli Ban]

Snopes reviewed internal documents from Anthropic that show they have bought millions of books to "destructively scan" them to train their AI model.

https://www.snopes.com/fact-check/ai-co ... are-books/

Ok as long as the book is not extremely rare or national heritage or whatever. No evidence that they're destroying actual rare books though, so the outrage is premature imho

That's the fun part: Anthropic, like all idiotic AI/tech bro idiots, will NEVER defend themselves.
Literally all any of the AI companies had to do for years was just EXPLAIN themselves, but they rarely if ever do.

I was first left flabbergasted by this with Sora 1, when it was clear OpenAI intended that as more of a world modeling experiment, but they let the entire internet rage at them for basically no reason, and then whined about it and how much they loved working with each other, and the entire time I was screaming at them "JUST. FUCKING. EXPLAIN. HOW. IT. WORKS."
And they never fucking did!
Even if no one buys it, just explain how the tech works and what you intend to do. More than enough people will give you the benefit of the doubt.

Anthropic will proceed to never explain why they're destructively scanning books; it's telling that the ENTIRE discourse around AI is driven entirely by detractors, and it's often those same detractors that have to debunk a lot of the claims because the companies themselves apparently have no PR team whatsoever.

Seriously, there's an almost suicidal lack of awareness or ability to read the room that took hold of the billionaire class over the past decade, and I can't explain it. Not even arrogance or a sense that everything will work out; they legitimately don't even seem attached to reality anymore. The whole "students booing CEOs selling AI" thing, I now think, ACTUALLY bewildered them.
Like, before I thought "They're powering through because they don't think they need to explain themselves, because they intend for this to all pay for itself in a few years and just need to ask for forgiveness then."
But now I'm not so sure; there's a part of me now increasingly feeling like if Dario and his team faced enough public pushback, they'd legitimately unironically be astounded that anyone would find what they're doing suspect, when literally the controversy would be mostly quashed if they explained just WHAT kind of books they were destructively scanning, almost none of which are in any way creative or artistic.
Instead, they let the rabble run rampant claiming "Anthropic is literally destroying history, they're erasing authors and creative expression itself"
This is provably false, a lot of the claims about generative AI are provably false
But I have to rely on either fascistic crypto-bros for that proof, or find anti-AI types that aren't utterly insane who nevertheless always say "Okay, so AI water usage isn't AS bad as claimed, BUT... [here's why AI is still evil]" or "Okay so the hysteria over destructive scanning isn't as scummy as we thought, BUT... [AI is still totally evil, I'm only slightly conceding here and will totally relent when the comments tell me to not concede]"

At this point, even if the AI companies did start explaining themselves, it's years too late, they already let the narrative escape them, but they don't seem to have any intent of doing so regardless

And it's just... like... beguiling? It's like watching a creature that utterly lacks any sense of self preservation, and you're increasingly left in a daze how it managed to survive long enough for you to even be able to see it.

It's why I just don't engage with AI discussion whatsoever unless I absolutely have to, and usually come down against generative AI but for AI in general (which to be fair is my actual opinion, at least until generative AI proves itself more). It's a clusterfuck, it's all hysterical, and the pro side has never done itself favors. The pro side has to rely on actual deeds and examples, and we've literally only just started seeing such in the past few DAYS. Better to just avoid the minefield altogether, because AI-bros are determined to step on every mine they can and act wide-eyed why they keep exploding

[firestar464]

I'm not so sure about "almost none of which are in any way creative or artistic" but otherwise great points

Maybe it might have something to do with litigation risk, because lawyers might advise companies to avoid explaining shit if lawsuits are forseeable, due to the fear of giving plaintiffs ammo?

[wjfox]

[firestar464]

People were asking if they prompted it properly, which is a fair ask. Minor changes can make all the difference lol