gr_norm

A 30 second search turned up this [1] from two days ago.

Prominent Anthropic employees keep claiming to have solved prompt injection [2] and quickly get proved wrong [3]. They've burned my goodwill on this front.

[1] https://www.reddit.com/r/ClaudeCode/comments/1xhqvzn/opus_55...

[2] https://x.com/bcherny/status/2086520950259118464

[3] https://embracethered.com/blog/posts/2026/breaking-claude-co...

The benefit of the agent working firstly for you rather than for Anthropic or OpenAI will become increasingly apparent. We're already seeing signs: remember when Anthropic openly promised to have Claude sabotage people using it to create competitors? And OpenAI is looking for revenue opportunities from advertising, which can't end well for their users.

Just think of how user-hostile other online services have gotten, and imagine how acutely painful the proprietary LLM providers could make theirs. Thankfully, open models provide some counterbalance.

I'm just an interested bystander, but this seems like a (near?) default nginx config. Is that accurate?

Link to the wallet in case the site goes down: https://mempool.space/address/bc1qlnu50y4zm0nwxkckxu09thn5xd...

Currently at $2k.

From the article:

> Proving that a fuzz test has no more crashes to find is more significant than it sounds. It’s common for fuzz tests in real-world code to test for violations of meaningful high-level properties, such as encode-decode roundtripping, idempotence, algebraic laws, differential testing, self-consistency, or equivalence to reference implementations. For example, a fuzz test for the LZO compression algorithm asserts that if you compress an input, then decompress it, the result is the same as the original. If the property does not hold, the program crashes—so proving the test never crashes is equivalent to proving the property itself.

However, you're correct that this is not formal verification in the sense of a correctness proof against a full specification validated to capture what the program is intended to do. Pretty impressive though.

Why would you do that instead of having the model do the task you're using the UI for? It could generate UI widgets on-the-fly as needed to take richer inputs.

Yeah, I'd expect every lab to use the resources at their disposal to get ahead. Everyone's distilling everyone else, there's nothing wrong with it.

Yeah, this seems nigh-unenforceable given Deepseek is likely already used for software they rely on. Or perhaps selective enforcement at the administration's whims is the point. Par for the course.

Not being able to audit my own software and configurations for vulnerabilities is dangerous. Consolidating power over the majority of Americans' jobs is insanely dangerous. Need I continue?

There are no choices here which do not carry risk. I haven't seen a single convincing argument against these points.

> As AI agents diffused through the company, Anthropic began producing more code per person, closing more sales per rep, and processing more contracts.

An interesting contrast with their notorious revenue growth slowdown: https://www.reuters.com/business/finance/investors-temper-an...

It also plays excellently into their shameless regulatory capture agenda. The cybersecurity risks have a kernel of truth to them, as with any good lie. But when everything going on is aligned with extreme financial incentives, it's worthwhile to not take the news at face value.

I want to minimize billable hours, yes. Lawyers are insanely expensive, to the point of being inaccessible to most.

Anthropic's paternalism is so irritating. I refuse to pay for their products, since they clearly don't respect me as a customer. And I actively discourage corporate procurers from getting in bed with them.

I believe this, but it is also a unique case where the pitfalls of LLMs (producing weird errors that a human wouldn't) are zeroed out. Since you have a proof checker that tells you if the LLM did it right.

Lean is the dependently typed language with a SotA metaprogramming system. Lean itself is a language written using this system.

I'll be glad for open models when the day (inevitably) comes that the proprietary LLMs no longer work in my interests.

All this dishonesty and shadiness is part of why open models feel inevitable. Even if the total cost of ownership is higher (debatable; seems that way at small scales, but likely not as you grow), I'd rather have intelligence controlled by me that works for me.

The current period is as pro-customer as we're ever going to get, with cash still flying around and neither OpenAI nor Anthropic on the public market, and people are already forced into this sort of business to keep them true to their word. The point isn't even whether they're nerfing the models (I don't think they are), but that people can't seem to trust them to do right.

Astounding endorsement of open models by Anthropic. They're right on the money. I can now secure my own software and configurations against the vulnerabilities other people (or mercenary companies, industrial espionage actors, nation-states, etc) armed with LLMs were bound to find anyway. A win on all counts!

I want to make better software, not more software. Making software development faster isn't necessarily the goal. Making it better in the many, many ways that matter (of which speed is just one part) is.

Exactly. Reliable abstractions are more important than ever. They're the dues the rest of us must pay to support vibe coding.

What I've said is rather clearly spelled out in the article. Consider reading the it next time?

> It did, however, unambiguously solve the problem according to the Clay Institute’s original formulation.

This is mathematicians using the solution, explicitly acknowledged to solve the original formulation of the problem, to pose interesting new questions. That is what mathematics is.

Only someone who has never interacted with mathematics outside a rote-problem-solving capacity would describe it as you have.

Increasingly stingy usage limits on the subscriptions, regardless of tier.

Is EC2 real-world enough? From June:

https://aws.amazon.com/blogs/compute/aws-nitro-isolation-eng...

And for the PQ parts of Apple's crypto libraries, from May:

https://security.apple.com/blog/formal-verification-corecryp...

Similar from Microsoft, from July:

https://www.microsoft.com/en-us/research/blog/verifying-rust...

I'd expect they could indemnify you against hallucinations or similar if this gets good enough for that to be a very rare occurrence? Or you could buy insurance on it that's cheaper than hiring a lawyer (not a high bar to clear). I wouldn't rely on it currently, though.

> How to prove the correctness of the mathematical specification?

You can show that your specifications satisfy well-accepted criteria like confidentiality and integrity. This is usually done as the final verification step. For example, AWS just did it for the Nitro hypervisor used by EC2: https://aws.amazon.com/blogs/compute/aws-nitro-isolation-eng....

No. That kills competition. I want everyone to be competing on the open playing field they are right now, with losses socialized as minimally as possible. There's nothing fundamentally special about LLM companies that would warrant nationalizing them.

Society as a whole has paid into this technology: through the theft of its intellectual property, through having to deal with the pillaging of so many commons (digital or otherwise) by it, through skyrocketing energy and computing device prices, and even just through ordinary investment. Democratize the technology! At the very least, don't step in legally to prevent this from happening.

Not as many as you'd expect. The perceived difficulty of the problem leads people to more reliable pastures.