Daniel Jeffries adds that such restrictions enable regulatory capture
There's a lot of sloppy thinking around open models. You can ban them and make it impossible for US companies to use them, but this won't stop A) global open model progress B) bad actors using them
So what exactly is gained by banning open models, including those from China?
Many users criticized bans and regulations targeting OpenAI models as brain-dead ego-stroking power grabs by officials, while agreeing such measures fail to stop global progress or bad actors.
No Digg Deeper questions have been answered for this story yet.
@natolambert What is gained is regulatory capture.
There's a lot of sloppy thinking around open models. You can ban them and make it impossible for US companies to use them, but this won't stop A) global open model progress B) bad actors using them
So what exactly is gained by banning open models, including those from China?
@natolambert To be clear, the "you can ban them and make it impossible for US companies" line is also backed up by a lot of sloppy thinking as well. But as you point out there's a more fundamental policy question.
@natolambert I'm just here to agree with you.
@natolambert Supply chain risk, that non-US models may have backdoors/sabotage capabilities. I think it's pretty unfounded but there we go.
@natolambert Alot of conspiracy theories around AI is surfaced on the internet. Maybe, they are watching it.
@natolambert economics for the US closed model companies i suppose - perhaps that's always the calculus
@bahradx say more? There's policy ambiguity but thats roughly the scope and the likely effect
Regulation and more money for the prison industry complex Nathan, I hate to say it, but look at the cannabis industry if you want a window into the possible future of distributed intelligence. Entire regulatory divisions are funded and the crushing wheel turns into accepted popular opinion. If it goes that far, its the bad roads/dictator theory. I am coming from a place that is 49th in education in the US, it is hard to see the current actions as anything other than a power grab with a "what are you gonna do about it" right behind. You know who wont help any of us? The people putting together committees to decide what models get released int he future. It is hard not to be negative, and I had codex scrub all negativity from my feed yesterday =p Trying times for people who really care and are not greedy.
@natolambert I agree. But sloppy thinking hasn't prevented the formulation of brain-dead regulation.
@natolambert I don’t think it’s a good thing but it would disincentivize funding for the companies building them, right?
@natolambert they are working on the de facto of banning these models before the de jure announced. im calling it.
@natolambert Market share.
That's the crux. It's all about the money.
@natolambert the other concern is the consistent pattern of frontier capabilities diffusing to the hardware already in circulation, and these prescribed risks being lab repertoire beyond some genius synthesis of theory.
@natolambert > what exactly is gained by banning open models
What exactly is gained by banning Mythos and GPT 5.6 and Fable?
You're dealing with a cabinet full of fragile-ego midwits in the White House. Neither logic nor reason is an asset in this conversation.
@natolambert This may kill many neocloud companies in the US and anthropic will get more power.
@natolambert Is just stroking the ego of some leeches in power
@natolambert honestly this is the part that never gets answered in the debate
people just skip from "ban" to "win" with no mechanism in between
@bahradx @natolambert Every single company that has advanced tax planning or any sensitive data will NOT be able to use public models because they are admissible in court. They will use local models. If they can’t use those models in US they will use them in subsidiaries in other countries