Critics compare Anthropic's lobbying to late-1990s proprietary UNIX tactics.
The whole reason Anthropic has been able to secure and retain talent is due to the narrative he engineered that they are the heroes and only they can be trusted to create and infer models.
Regulatory capture and banning OS models is a natural byproduct of this mental bias.
The number 1 public enemy of open-source. I'm glad that many are starting to see who Dario really is.
Some users praise Anthropic's strong models and enterprise products while others accuse its CEO of regulatory capture to block open-source competition driven by greed and a messiah complex.
No Digg Deeper questions have been answered for this story yet.
Like many of you, the question of why Anthropic is acting so batshit keeps rolling around in my mind.
Have finally come to a conclusion.
It’s existential risk alright. For themselves, not humanity.
Why? Let’s talk open source history.
If you weren’t around in the late 90s/early 00s it’s hard to describe how thoroughly Linux wrecked an entire industry of UNIX hardware and software vendors, practically overnight.
Sun. SGI. HP. SCO. Countless smaller vendors.
All saw that business evaporate, taken by an open source OS running on commodity hardware.
There was no coming back. Some tried to embrace Linux on their hardware, didn’t really work.
This time, there’s not even a hardware play to fall back on for Anthropic. To add insult to injury, the most popular hardware vendor is making open source models and giving them away day one.
In the earliest days of Linux they ignored it, but the old guard remembers what happened.
They did eventually wake up. Tried dirty tricks.
Embrace, Extend, Extinguish to try and ruin open standards on the web and elsewhere.
FUD. Propaganda to spread fear, uncertainty, and doubt.
None of it worked.
Today they have a new weapon. Regulatory capture. The models are scary. More FUD. They learned.
They try to get the government worked up about a race with their adversary. Limit China’s labs from distilling their models.
But they don’t care about that. They can’t even sell their product in China.
It’s open source that they don’t want to compete with.
They tell on themselves by being the only major lab to not release an open source model.
OpenAI, SpaceX, they have chips now. Datacenters. They’re hedged. Anthropic does not. Is not.
So they angle for a lockdown. Today it’s frontier. Tomorrow they ask for everything.
Tell them they can’t have it.
Can’t stop the signal.
@archived_videos Appreciate the reply, I think if it was just the former the money behind them would be losing their shit.
@archived_videos Agree, fantastic products. Same was true for the UNIX vendors. I love that hardware.
Doesn’t matter, if allowed to flourish open source wins every time.
@mov_axbx I think it's more "they're weird and don't share your values" than "they share your values and lie to try to capture an industry"
@mov_axbx It's like, nobody seems to be doing "I'm openAI except there are way more models to choose from and they're open source models", you have openrouter but they don't have apps, opencode/pi are not useful for non technical people, and then you need zero data retention policies
@mov_axbx Yeah, currently I think what's missing is an easy way for companies to get the equivalent of Claude or ChatGPT, so cowork, artifacts, shared skills, shared MCPs, at a kind of knowable price
@mov_axbx The thing is, they still have great models, great products, great enterprise penetration. Yeah they're weirdos but they actually deliver the goods. Lots of insane people in SF, insane people that can also reach profitability at this scale, more rare
@archived_videos That’s one extrapolation.
You can also say they don’t view it as strong of a threat, trusting in their ability to adapt.
It was also a fairly weak model and they definitely earned some points with those concerned about the national security implications of foreign models.
@mov_axbx Following that train of thought you end up with stuff like "openai released an open source model only to not be accused of regulatory capture"
@archived_videos I think you’re right and I bet things evolve in the way you suggest as compute becomes more abundant.
@mov_axbx @archived_videos “Allowed to flourish”. By that do you mean elimination of free market mechanisms, prevention of monopolies, or something else?
@mov_axbx I hope they will! I'm happy paying ~$20 a month to one of the big labs, but as soon as I need API usage I go look at deepseek, and sometimes have a weird mix of github actions, models through the api, opencode in github actions, etc
@mov_axbx Also Cloudflare and Stripe have talked about what they've built, Cloudflare for code review and Stripe for ticket implementation, and it's those kind of systems based on open models. But it means you need an internal "devs for devs" team
@beffjezos It’s pitch-perfect woke in its irresistible appeal. A full measure of narcissism plus moral Pokémon points.
@mov_axbx @yacineMTB My reply history is littered with "they have no moat" for 6+ months.
@mov_axbx Yup I'm following you.
@beffjezos Enron level
@Shine_holon @archived_videos I mean not making it illegal through regulation driven by fear tactics.
@beffjezos All three labs hire on narrative (safety / constitutional / open). the question is which survives when reality bends. Anthropic's is the most vulnerable
@beffjezos Greed + messiah complex = WHAT COULD POSSIBLY GO WRONG?
