ronsor

I understand preferring Claude models in the abstract, but as a product? The random restrictions have significantly reduced their usability.

We had 2 major threads in the past 48 hours complaining about this:

* https://news.ycombinator.com/item?id=50319358 (AI/ML work with large models)

* https://news.ycombinator.com/item?id=50320255 (reverse engineering)

The threads are full of upset comments!

The only non-OAI options are open models and I guess Gemini 4 now.

There are later dumps, but they aren't quite public

The point of storyboarding is to determine what you want to make. I guess you can consider it part of the "prompt."

Not something that can be automated away assuming you still have preferences at all.

> many have memory corruption bugs and incorrect anti-tampering code requiring our per-app exploit protection compatibility mode toggle

Ironic for apps that are often so paranoid about "security" and "hardware attestation"

Reality has been stranger than South Park for a while now.

anti-AI scraping is often anti-privacy in practice, as any effective implementation requires invasive browser inspection to detect scrapers.

More like Monero, but yes.

They're down, slow, and congested, but spend 1/10000th that companies shackled by big cloud do.

Are you saying that legitimate businesses cannot find something better for 1/100th the price instead?

Three months ago is about when I gave up on trying to use their newer models for almost any task without getting blocked.

But EC never seemed to cause trouble somehow.

Four more organizations needed to secure their systems better, I suppose.

Facetious maybe, but in six months, foreign crooks will be producing these incidents on purpose.

I know Anthropic models have a lot of insane, paranoid filters, but I don't remember them blocking cryptography work last I checked (some 3 months ago now).

> should they choose to exercise it

Until other networks choose to exorcise them and their control with it. If you play unfairly often enough, you get depeered.

LPEs are usually used to escalate to root, but in many of them, the kernel itself is compromised. There's just little reason to do stuff in kernel space when you can get root and load kernel modules anyway.

All of this will defeated by one of the now regular Linux LPEs anyway.

Because card readers brick themselves if opened or tampered with. If Ledger designed their hardware the same, this attack would not be possible.

OpenAI and Sam Altman are literally backing a nuclear fusion power startup.

That's just a YouTube embed

Ah, yes, the mighty EULA. Bad actors always kowtow to these.

www.x would be owned by Elon. Don't give him ideas.

Steam DRM is easy to bypass, but the convenience of one-click purchasing cannot be bypassed.

Until a major manufacturer that isn't Meta shows up.

I wish we had fewer products then.

So should everyone buy 3D printers before the government lockdown comes into effect?

Other than being in Rust, how does this compare to a CPU-only build of llama.cpp, which is both easy and already well-supported?

I didn't know they changed the page until I saw this article. Interesting.

Minecraft is the exception that proves the rule, if you ask me

VHS has an effective resolution of ~240p anyway.

> Authors: Ilya Sutskever, Dario Amodei, Sam Altman

Only one remains at OpenAI.

Anthropic is trying so hard to "crack down" on distillation that they're ruining their own product. I do not know why.

They already stopped. That's why the service quality declined.