House panels probe Airbnb, Anysphere over use of Chinese AI models
nextgov.comIs this an effort by OpenAI/Anthropic to force US companies to use their more expensive closed-weight models? I don't see the problem with US companies using Chinese models if they are hosted in the US.
This is Anthropic's plan.
To stop us from using powerful open-weight models from running on your machine and they want you to pay for their expensive and over-priced models.
"This is Disney's plan," He uttered, watching in disbelief as his Cinderella Blu-Ray rip was removed from Google Drive for copyright infringement.
This is the US government's plan. OpenAI, Anthropic and X are just obedient lapdogs that the DoJ can rely on to fight for the worth of their intellectual property.
Except that at least 2 out of 3 of those companies have released open-weight models.
Can you name one open weight LLM that Anthropic has publicly released?
OpenAI and X's open-weight LLMs are terrible. Both of them deliberately nerfed their releases to prevent competition with actual SOTA models.
I consider all three companies to be demonstrably hostile to local inference. Anthropic at least doesn't lie about it.
> OpenAI and X's open-weight LLMs are terrible.
They are open weight models that they have released and we know they are not the only ones out there to use and it is better than none.
> I consider all three companies to be demonstrably hostile to local inference.
Google isn't hostile to local inference and also release open-weight models for local inference.
> Anthropic at least doesn't lie about it.
You literally get nothing for defending Anthropic who doesn't care about you and continues to lie to you about "safety" because the truth is local models directly compete against them.
> we know they are not the only ones out there to use and it is better than none.
Oh, my apologies. It's not fair to compare shoddy upstarts like OpenAI and X to Deepseek or Qwen or GLM or Kimi or MiniMax for not releasing their best model weights as soon as they're out. I'm just not appreciating all of what OSS-120b and Grok 2.5 have to offer to the homelabbing community.
> Google isn't hostile to local inference
I didn't include Google among them. "all three" meant X, OpenAI and Anthropic.
> You literally get nothing for defending Anthropic
My quote was not "Anthropic at least doesn't lie" full stop, was it? They don't lie about being a frontier local model provider.
OpenAI hasn't released a genuinely boundary-pushing model since GPT-2. That's a contemptible output for a trillion-dollar research organization, say what you will about Anthropic's corporate policy.