News for hackers. Stuff that matters.
Re:Fuck Your AI Safeguards by hintymad on Thursday November 19, 2026 @03:52PM (#50566192)
> The safeguards are the future of closed AI

What is our solution? Relying on Chinese open-weight models will be unreliable, to say the least. No one will invest another frontier lab like Anthropic or OAI, let alone one that will release the weights. Unless the technology improves like how workstations outcompete mainframes so the open-source community can build our own models, I fail to see how we change the situation.

Re:Fuck Your AI Safeguards by octoberfranklin on Thursday November 19, 2026 @04:46PM (#50566906)
I've found Chinese open-weight models to be by far the most reliable models.
Re:Fuck Your AI Safeguards by hintymad on Thursday November 19, 2026 @05:09PM (#50567151)
By "unreliable" I mean that we don't know when the US government or the Chinese government will restrict the access to the models, or the Chinese companies will stop release the weights when such act no long makes business sense to them.
Re:Fuck Your AI Safeguards by ImPostingOnHN on Thursday November 19, 2026 @06:18PM (#50567964)
If that happens, the models on hard drives still work as well as they always have
Re:Fuck Your AI Safeguards by docshock on Friday November 20, 2026 @12:23AM (#50570491)
I've found Chinese open-weight models not to be truthful. There are planted falsehoods. Hard pass.

NVIDIA, on the other hand, released a huge training corpus and open models up to the 500B class.

Re:Fuck Your AI Safeguards by aembleton on Friday November 20, 2026 @09:14AM (#50574942)
> I've found Chinese open-weight models not to be truthful.

What do you mean? I've only used them for writing code and they seem to work well enough for that.

Re:Fuck Your AI Safeguards by docshock on Friday November 20, 2026 @09:46AM (#50575362)
Ask your Chinese model to explain Chinese geopolitics, and it will parrot falsehoods.

Now ask yourself: What else have I not inspected in this model?

The Chinese models have no technical advantage over open-source, open-weights models of known provenance.