[neuronexmachina]

Based on this, it seems like the Trump admin would have targeted them even without the "scaremongering":

> To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws. Our understanding is that one potential jailbreak was shared with the government. We have reviewed the report and validated that the level of capability displayed there is widely available from other models (including OpenAI’s GPT-5.5), and is used every day by the defenders who keep systems safe.

[zmmmmm]

> the level of capability displayed there is widely available from other models

Is this Dario leveraging it into a ban on open models?

[PlasmaPower]

No, he specifically gave a proprietary OpenAI model as an example (unless you meant OpenAI models instead of open source models)

[nullbio]

Anthropic is almost solely responsible for the fear narrative around AI at this point. It has been their culture since the beginning, strongly pushing this into the zeitgeist at every opportunity, releasing bogus papers to frame things as highly dangerous and that their AI is a conscious sentient being.

Step 1: "OMG, the AI hacked a researcher eating a sandwhich in the park!"

Step 2: Journalists use that great clickbait to generate profit, which generates publicity for Anthropic

Step 3: Rinse repeat

If the threat of LLMs was treated relative to the actual capabilities of them, and we weren't all being lied to by Anthropic and their army of millions of social media bots and backing media companies and mouthpieces, we'd be going in a much healthier direction. Working out the kinks/supply chain risks and developing sound, long-term countermeasures to the ACTUAL risks.

The only threat to the world is if progress is not open-sourced, democratized and in lockstep with capability. The moment it becomes a scenario of: Only a small group get access to frontier intelligence, is when it gives that small group power over everyone else in the world, and wildly increases the risk of a nuclear level event that WILL be exploited eventually - as the divide between the haves and the have nots accelerates in an exponential fashion. Bad AI is countered with an abundance of good AI that has been used to stay ahead of bad AI. The moment your bad AI outpowers the army of good AI it is game over for humanity. The strength of open-source and open-access AI is the difference between humanities permanent enslavement or extinction versus a prosperous future.

It doesn't help that most of the employees at Anthropic have willingly sold their souls out of short term greed and gaslight themselves into thinking that they're actually doing the right thing to justify their own greed to themselves, while building up an echo-chamber and culture of feel good lies within the company so they can sleep at night, and pat each other on the back. They go along with this because they get paid massive chunks of money from Anthropic, and their shares will be worth more money if Anthropic can swallow the worlds economy at the expense and enslavement of everyone else. What good is that money when you have to sell out humanity in the progress though. You, at Anthropic, is that the legacy you want to leave?

People need to start calling this out before it's far too late. If you work at Anthropic - time to start talking to your colleagues in an honest manner.

[dannyw]

It seems like this might not have happened if Anthropic didn't institute the cyber blocks as broadly to also cover pure-defensive use cases?

Because this wouldn't be considered a jailbreak with any other model; which would just do the request.

[Eridrus]

The difference between OpenAI & Anthropic is that OpenAI didn't do multiple big media pushes about how their models are so scary and dangerous.

OpenAI's models are very good, they have refusals + a government ID verification story for cyber access (I don't think they prevent non-US nationals, but I don't know this). What they don't have is Project Glasswing and all the hand wringing about how they're going to end the world in public.

I hope Anthropic pulls their head out of their ass and just starts acting like a normal company.

[jazzyjackson]

You’re telling me this testimony isn’t sincere marketing for how revolutionary and dangerous his product will become?

  OpenAI CEO Sam Altman testifies at Senate artificial intelligence hearing | full video“ (2023)

"My worst fears, are that we cause significant - we the field, the technology, the industry - cause significant harm to the world...If this technology goes wrong, it can go quite wrong and we want to be vocal about that."

https://youtu.be/Pn-W41hC764

[datadrivenangel]

OpenAI did this back in 2024 several times.

[Eridrus]

I think Anthropic went a lot further with the marketing around Mythos/Glasswing.

OpenAI talked a lot about potential future risks. Anthropic went around saying "this model is too dangerous to give to people at all".

[ceejayoz]

That’s a safe assumption, considering they tried it a few months back too.

https://www.npr.org/2026/03/09/nx-s1-5742548/anthropic-penta...