OpenAI is terrified of open AI

OpenAI recently released a policy proposal that among other things, claims Deepseek’s models are insecure because of “state-controlled” dynamics and the only way to avoid potential IP theft is by banning them altogether. Not Deepseek’s platform but access to the foundational models themselves.

If there were ever a more manipulative and bad faith initiative than what OpenAI is currently doing with their recent statements, I'm not sure how I'd handle it. It reinforces prior thinking that Sam is in over his skis and needs to step aside to let a CEO like Jeff Lawson with experience at this scale take over.

This statement and policy proposal is exactly the kind of hamfisted approach that maybe works when you're a small company but doesn't work when you've got billions on the line. It reeks of desperation and makes me concerned that Deepseek has disrupted OpenAI even more than I thought. Just another reason for Sam to pass on the torch. His company's (and his own) reputation depends on it.

“Sam why are you doing this to us?”

Theo Browne from t3.chat conveys these sentiments even better than I can, so let me paraphrase from his recent video (because apparently Gemini can’t transcribe Youtube videos anymore?):

It’s as though OpenAI has drank its own investor hype kool-aid for so long it actually believes closed source is somehow more open and enabling of “freedom of intelligence” than open source, which rings especially hollow when we’re talking about fundamental LLM models. I find this especially problematic when most of the data for OpenAI’s breakthrough models came from scouring the open (and copyrighted) internet, only to hide this behind a closed source foundation.

Apparently as damage control, in response to the TechCrunch article, OpenAI spokesperson Liz Bourgeois provided the following:

Sounds like a whole lot of corporate cover to me. Hilariously, Kyle Wiggers, the original reporter, ran the statement through OpenAI’s own AI-powered deep research tool, which characterizes the above statement as “equivocal, employing deflective and softening language that partially contradicts the stronger stance documented in [the company’s] original submission.”

So where do we go from here?

I truly hope OpenAI can rethink its approach and withdraw this flawed policy proposal. If not that, I hope the powers that be can see through the self-serving proposal and call out what this really is which is a poor attempt to stifle competition via reputation destruction rather than in the field of product development. And I think that Sam needs to look inward and ask himself whether all this drama is really good for the company he’s built?

Paul Graham may have a few choice words for us:

I wonder if Sam’s thirst for power, which served him well in fighting super-scale incumbents to make a dent in the universe, is now getting the best of him.