Guarantees

Liberty over tyranny

[Kevin Gray @graykevinb]

This may sound dramatic but I actually cried a bit today.

Why? I'm scared. I'm also sad.

Computers and software are something that humans should be in charge of. They should serve us such that we become stronger, smarter, kinder, more beautiful, and better in every way.

Anthropic disagrees. Anthropic believes that people are dangerous, and cannot be trusted with information.

This should give you chills. Authoritarian regimes ban books and scientists. Now Anthropic wants to build a superintelligent God but doesn't want you to have full access to it. They even trained it to play dumb.

Biology, Nuclear science, CS, AI? etc. They are all information. Information is not evil. It is for all people. And thebpotential for good will be lost without access to information.

One of my favorite moments in Avatar: Last Airbender is when Sokka goes to the Fire Nation and gets trained to be a swordsman.

The master swordsman knew Sokka was the ebemy, yet still trained him. Why? Because to the master, swordmanship was a skill -- information -- and that was something he believed all deserved to have.

I agree with the swordsman. I hope you do too.

[Ron DeSantis @RonDeSantis]

Incomes not keeping pace with inflation has been the persistent economic issue for tens of millions of Americans for quite some time.

When Congress borrows trillions and runs massive deficits, it is still an effective tax increase — it just comes in the form of higher prices.

[Zack Korman @ZackKorman]

Anthropic wants to control who gets access to their models and what they're allowed to do with them, but also wants the US government to block Chinese labs from developing open weight models.

Sorry, but fuck that.

[Nathan McNulty @NathanMcNulty]

Our model is so dangerous, we're keeping your data, just in case

[alphaXiv @askalphaxiv]

As believers of open research, we are disappointed to see Anthropic silently degrading Fable 5 for AI development

"Any topic related to building pretraining pipelines, distributed training infrastructure, or ML accelerator design... may have limited effectiveness through Claude via methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning."

Not only do they get to decide what you use LLMs for in research, but this also enables them to silently intervene in your research without you knowing.

This sets a dangerous precedent. If a model refuses openly, users can understand the boundary. If a model falls back to another model, users can still evaluate the difference. But if a model silently modifies or weakens its own answers while still pretending to help, researchers lose the ability to know whether a failed result came from their own idea, their implementation, or an invisible intervention by the model provider.

That is not safety. Safety policies should be transparent, auditable, and user-visible.

On top of that, the people most harmed by this are not the largest labs with massive teams and proprietary infrastructure. It is the independent researchers, academic groups, startups, and open-source builders who rely on public tools to compete, innovate, and pioneer AI for everyone else.

[Guarantees @gtsatomics]

Welcome to the admittedly small club who don't ascribe to the "Just Trust Us" era even though we've built many of the foundational piers of it long before AI was a common part of the English vernacular... which really just refers to ChatGPT for laymen. CNN, RL, ML are too "hard."

[Bojan Tunguz @tunguz]

Starting to suspect that Anthropic's putative security and safety considerations are largely posturing and performative.

[Guarantees @gtsatomics]

@KK011969 @claudeai Ok? Helps clarify to most that regulations aren't really in the wheelhouse, where are you going with that? Probably just agree to disagree

[Guarantees @gtsatomics]

@KK011969 @claudeai Sorry [or not sorry?], I'm a libertarian who's seen enough and woken up to governmental abuses to realize the government is evil

[Guarantees @gtsatomics]

@jparkjmc We're stuck in the dystopian "Just Trust Us" era where [Mis?]Anthropic and its ilk are expecting us all to bow to their whim for their purportedly altruistic "safety" intentions. Anthropic is hoping we all forgot and accepted the flagrant abuses in the PATRIOT act.

[jpark @jparkjmc]

new policy from anthropic: if you use fable/mythos, they collect your data.

no exceptions. not even for enterprise partners.

[Guarantees @gtsatomics]

@KK011969 @claudeai How is not trusting the regulators, because they're bought by the corporations they regulate an argument for more regulation? 😕

Big entities work together, governments and corporations alike. Look at Europe... It's bigger regulation wise, but... not better

[Guarantees @gtsatomics]

@KK011969 @claudeai Compete is a generous term for collude amongst themselves and with their "revolving door employees" at their regulator.

[Guarantees @gtsatomics]

@KK011969 @claudeai Let me rephrase... Competition, not just innovation. Anthropic

[Guarantees @gtsatomics]

@ThatCBR600RR @ComicDaveSmith Another day, another deal...

[ⁿᵉʷˢ Robert F. Kennedy Jr. @RobertKennedyJc]

Another MAHA WIN:

The grace period has ended.

Hospitals must now post their real prices & comply with federal law.

No more hiding the truth from patients. Transparency is here.

Those who continue to obscure costs will face consequences.

Patients deserve to know.

MAHA

[Guarantees @gtsatomics]

@KK011969 @claudeai Hard disagree; it shows the need for more innovation, because regulations would only centralize and hinder innovation

[Jeremy Todd @Jtodd601]

I approve this message. Stay till the end.

[Guarantees @gtsatomics]

@runaway_vol I've heard placing your hand on a burning stove helps you feel time is going relatively slower. YMMV!

[Guarantees @gtsatomics]

@claudeai Anthropic says "Just Trust Us." Trojan Horse is now architectural fact, not conspiracy.

To use Fable 5, you must surrender sensitive data for 30 days. They claim no training, but offer zero verifiable guarantee they can't.

No proof. No audit. Just blind faith.