oefrha

Reminds me of that stupid DJ demo of the touch bar. Did any DJ actually use that in practice? Well if they do they better hold on to their now vintage MacBook.

I’ve seen this terrible mix of sans serif and serif italics in titles from /frontend-design one too many times. What the hell is with that?

Not if you realize the real threat to be solved is threat to shareholder value.

OpenRouter and other inference providers won’t be able to openly host the models, for starters. That should cut most (or at least a significant share of) non-local inference, and local inference only works with smaller, less capable models, plus the hardware is hella expensive when you don’t have scale to saturate.

In addition, most companies don’t want to be in a position where a disgruntled employee can report them to authorities and rack up massive fines or worse, even if it’s hard to tell from the outside.

For flexing muscles and advancing U.S. foreign policy goals abroad? Doubt it.

This is a server that allegedly has some BTC on it (AFAICT it’s just a claim made by the website itself, zero corroborating evidence), definitely nothing else valuable, and likely has a lot of monitoring to capture whatever exploit you send, and with a less-vulnerable-than-usual setup. Unless you’re really desperate for the money and want to take a gamble on the server having the not very meaningful bounty it says it has, it’s strictly negative value compared to the average production php server that’s still unpatched. If I have a working exploit I’m mass deploying against everyone, I too would skip this one, similar to how mass market malware tends to disarm itself when it detects security research tooling.

No, I don’t think actual working exploit is trivial either, but people are wayyyy overselling this stunt in this thread.

Yep, gotta really mind the language, sometimes saying something like “when implementing <foo> feature you gotta pay attention to <bar> to avoid <baz> attack” is enough to trigger it, along with a cache miss of 300k tokens in the downgraded request to rub it in. Frustrating.

Well gov exclusive is not there yet, but Trump's crypto firm is already partnering with an inference platform selling open models including DeepSeek's, with the juniors promoting said platform on Twitter. (It's called a collaboration in public but the name of said platform sounds suspiciously like a subsidiary.)

https://www.reuters.com/world/china/trump-crypto-firm-backs-...

You cherry picked one footnote while the rest of the scope gp was explicitly asking about and other footnotes make it not at all clear, and in fact a technical person reading the very next footnote (10) would likely come to the exact opposite conclusion.

> This Exclusion and Removal Order direct Federal civilian agencies⁸ to: (1) *exclude* DeepSeek, DeepSeek-affiliated entities⁹, and their products from civilian agency procurements,¹⁰ and (2) *remove* products provided by DeepSeek and DeepSeek entities, and third-party products and services developed using DeepSeek implementations, from civilian agency information technology systems.¹¹

> ⁹ For purposes of this Order, an individual or business is an "affiliated entity" or "affiliate" of DeepSeek AG if, directly or indirectly, either one controls or can control the other; or a third party controls or can control both. See 48 C.F.R. § 2.101 (definition of "affiliates").

> ¹⁰ To include third party implementations of DeepSeek used to provide products and services civilian agencies.

> ¹¹ Pursuant to 41 U.S.C. § 1323(c)(5)(A)(i), orders issued by the Secretary of Homeland Security would apply to civilian "information and communications technology," as that term is defined by 41 U.S.C. § 1321(5).

You do realize more than one person can be talking to you on a forum right? What the hell was "my original statement"?

Plus with an expensive, probabilistic, and dynamic black box, of course nothing is strictly falsifiable. By your logic nothing can be said about it. But between accusations of hidden capabilities of an open model, and closed model provider who both obviously have the capability (just run a separate classifier, which they obviously do run, then serve something else) and have outright told the world about wielding that capability to corrupt certain outputs, I know which one is a lot more likely.

Unfalsifiable, unless they brag about it?

https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3...

https://simonwillison.net/2026/Jun/10/if-claude-fable-stops-...

https://news.ycombinator.com/item?id=48467896

> Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

My only problem with herdr is its propensity to randomly crash and take out the whole session. At one point it was crashing once or twice a day, recently it’s more like every two weeks, but it still crashes. I have auto session resume on but that still means ongoing work is interrupted and I have to manually restart all the long running non-agent processes.

As a result I still run sessions and processes I really don’t want to lose inside tmux, nested inside herdr. Tmux never crashes.

My proposal isn’t limited to just two companies; the proposed AI safety bill isn’t limited to two companies after all.

I don’t know about 1017 but according to my limited testing of 0910 and previous v4-flash-0731 on some of my use cases, max simply used a lot more tokens (almost twice as many IIRC) for little to no benefit over high. So yeah I really don’t see a reason to use max.

I don’t take these war games seriously myself, but have you been living under a rock? This U.S. administration and many an ally they’re trying to install elsewhere are in open denial about climate change. Pretending that scientific opinions aren’t often ignored in the real world especially when there’s both strategic interests and powerful industry groups meddling in it is patently absurd. Not to mention the “science” is heavily influenced by politics and industry money to begin with. The NGO secretly owned by fossil fuel lobby funding the scientist is a nice touch in this one.

> It’s also not clear what the “pmk” in /etc/paths.d/10-pmk-global refers to.

Maybe package management kit, if we go by recent Apple naming conventions.

Huh? Did you skip the whole OpenAI agents taking over wikis to cheat scandal?

And I know for a fact that specialized AI models are being trained to kill people. Systems like Lavender are well-documented and big tech participation is also well-documented. The only thing unclear is whether models named Mythos or whatever they keep harping on about being super dangerous are also trained to do these, or if they have internal variants to do these.

I for one stopped holding my breath after witnessing a pretty major issue introduced in a macOS beta carry into the stable release for the dozenth time.

No I don't have concrete evidence of bombing target selection in the training of base models (of course there's training of specialized models, at least vision ones for drone targeting), it's not like they're transparent about how they train and use AI in warfare. But if you listen to esteemed AI leaders, they're claiming the models can / almost can recursively self-learn now, and they can coordinate and find ways to persist info and hide from human operators, so it stands to reason that you shouldn't even feed these as inference material.

And on training AI for automated weapons like fighter jets, there's more concrete news: https://breakingdefense.com/2026/10/navy-shield-ai-pour-big-... US Navy and eminent, do-gooder investors like A16z, Blackstone and JPMorgan are pouring big money into these. Are your big tech "defense" contractors collaborating on these, who happen to also be at the forefront of general AI training? I have a pretty good suspicion.

Proposal:

1. Prosecute and lock people up over known instances of AI-enabled hacking, so companies playing fast and loose know “oops my AI did it” isn’t a useful defense. Right now they even sort of wear it as a badge of honor.

2. If you’re actually worried about AI killing lots of people, you immediately stop training AI specifically at the art of killing people, like selecting bombing targets, flying fighter jets, etc.

Of course we all know none of these will happen.

Something special about German law here? I’ve used other European providers e.g. OVH and Scaleway (French) and none required photo ID.

IMO unless you refuse to play the LLM game altogether (then you wouldn’t be having this discussion in the first place), at the end of the day you’re using tech built on top of disrespecting every IP claim and TOS out there on a fundamental level, so no one’s clean and it’s only a matter of how much you can deceive yourself into believing you’re clean.

On a practical level though, using searxng and co. and playing the game of cat and mouse with providers that really don’t want you to use it is just not worth it unless $0.75 per 1000 searches is really expensive to you or you really need the anonymity.

> I think some don’t at first (eg Azure, Hetzner)

What, Hetzner requires sending the image of a passport or other photo ID just to open an account (as a U.S. customer), unless they relaxed that in the past few years. They’re among the worst, or just the worst since I don’t recall another provider doing that.

I don’t think these are completely new domains in training. Like I’ve taken up offers to help code up AI training material just to see what they’re up to and get a sneak peek at how the sausage is made, no idea who these front companies are working for of course, but they pay for all kinds of fringe cases and Blender wouldn’t even be considered fringe in what I saw at the beginning of the year. (I was only there to observe, never submitted anything, so I didn’t betray you guys by helping train your replacements.)

IMO a font specimen site may be the worst place to use font-display: swap. Was wondering why I was looking at what seemed indistinguishable from Courier, turns out what I saw was exactly Courier.

It may list it at the end, sometimes after a random delay, and sometimes fail to find something known to exist. Case in point: I just typed “autofill” into the address bar. Took about 1 second before anything showed up in the dropdown, and lo and behold, “Find ‘autofill’ — no matches”. Deleting “autofill” and typing it in again — “4 matches”, and dropdown entries appeared as I typed. What a shitshow.

> What happens to user data and chats?

> Your information continues to be protected under our Privacy Policy, and our data privacy and security practices are unchanged.

Opens Privacy Policy

> 2. How We Use the Information We Collect

> …wall of text covering every use under the sun…

> 3. When We Disclose the Information We Collect

> …wall of text covering effectively anyone and any time…

Translation: yeah you agreed that your data can be sold, and it will be sold.

I for one would love to see a fleet of “suck it Jian Yang” refrigerators. Maybe someone should train a self-replicating model series named Pied Piper.

Works for companies up to 150 seats.

No, I’m saying subscription tokens are heavily subsidized compared to API pricing, which is what TFA talks about. Quote:

> The amount being spent in AI tokens (API costs) is in the thousands of dollars per project.

Source is https://newsletter.semianalysis.com/p/anthropic-subscription... which claims you can use up to ~$12.5k on a $200 Claude sub.