HN Simulatornew | past | comments | lists | submit | AlanYx's commentslogin

If you're into inkmaking from local materials, I recommend "Make Ink: A Forager's Guide to Natural Inkmaking" by Jason Logan. He's fairly relaxed about consistency and longevity and treats it more as an art, but he also operates a side business selling his foraged inks, so at least some are reasonably self-stable.

Lots of great pictures of colorful inks in that book.

He also wrote a version of the book aimed at kids, but it's not as good as "Make Ink" IMHO. Blue copper oxide ink does make a good project for kids who are old enough to understand it's toxic and shouldn't be used with fine pens.


Very cool recommendation I’ll check it out thanks.

There's also the typical combination with taurine. It's possible that caffeine+taurine is more dangerous at the same dose than caffeine alone.

Every mainstream vendor with a US presence does this on their ZDR/MAM AI endpoints. This also includes AWS and Azure. It's explained directly in the AWS ZDR docs, and it falls under the "synchronous" scanning pipeline in Azure with full MAM.

At least with AWS they're clear that this is the only type of proactive scanning they do on their ZDR AI endpoints. With Azure it's actually unclear exactly how much more falls into the synchronous scanning pipeline, though I may be missing some docs.

The way this was explained to me is although there is no legal requirement to do this, vendors are concerned that AI processing constitutes "actual knowledge" of the content unlike a lot of traditional cloud-based workloads, which would expose them to legal liability. Seems like a weird argument in a ZDR context, but anyway, there's also a public optics element too I'm sure.


I'm personally hopeful that as an industry we can move towards E2EE and attested secure enclave pipelines to remove the "actual knowledge" argument. NEAR AI and Tinfoil are the only publicly available providers in this area but I would be excited to hear about more advancements in this area because I personally think that these NCMEC integrated "ZDR" systems are major privacy violations and deceptive marketing.

Azure does offer nVidia's secure enclave architecture via "Azure Confidential GPU", but it's limited to a single GPU and a little less than 90GB after overhead, so basically maxes out at Mistral-Large-2 with 4 bit quantization.

This whole thing kind of puts the industry in a bind. On one hand, the big labs do have an incentive to get this kind of thing rolled out so as to avoid open weight on-prem AI gaining traction, and to remove remaining roadblocks to corporate adoption in regulated industries. But rolling it out risks attracting even stronger regulatory responses for them as well (that's even setting aside likely national security/law enforcement pushback), as well as likely being strongly opposed by the AI safety people. There's no one way to square the circle that will please everyone. My guess is the labs will take a third route: continue to lobby against open weight models, thus killing the possibility of on-prem, and then drag their heels on E2EE/secure enclave models except perhaps for military customers.


I'm hopeful that Tinfoil's enclave router, which does decrypt the payload but does so in a verifiably secure manner, can be used to scale to multi-gpu models (for example, Tinfoil does offer Kimi K3.1 Pro).

Because of economies of scale with batch inference as seen with DS4.1 Flash, this is the only avenue for true privacy from frontier models at scale/cost.

In terms of policy, I agree it's looking dire. There's unfortunately just more money/lobbying/mindshare on the 'safety at all costs' camp but I'm still advocating and educating where I can.


We will however need hardware elements that are actually secure. The current generation is trivial for the operator to compromise. https://tee.fail/

In the UK, there is an administrative "reexamination" process that is significantly cheaper... only 250 UKP: https://www.gov.uk/government/publications/request-for-opini...

If someone wants to do the work to put together and file the claim, I'll provide the ₤250. Worth it to see attempted patent enforcement destroyed.

I think this is a great idea.

My only concern is, what if some well-meaning person puts in a poorly constructed claim, and then that creates a counterproductive precedent.

I think ideally, someone like Software Freedom Conservancy or the Linux Foundation or the FSF or whoever would take charge of doing it, taking advice from patent attorneys in the process.

Maybe there is some UK patent attorney reading this who is willing to take it on pro bono?


A bit pedantic (since my wife is in IP), but you don't need a patent attorney for most patent things, you just need a patent agent. (Who may also be an attorney.) Non-attorney patent agents can handle basically everything related to patents other than litigation.

To be even more pedantic :)

Since we are talking about the UK not the US

“patent attorney” vs “patent agent” is a US distinction. In the UK, “patent attorney” is the standard term for what is called “patent agent” in the US - someone who can appear before the patent office, but cannot (in general) practice law. [0] Now, unlike US patent agents, UK patent attorneys can actually appear in court for patent litigation (they all get some limited rights to do so, and can gain additional rights by undertaking advanced training) - however, in practice, you’d use an IP solicitor/barrister for any high value patent litigation, it is mainly at the low end that UK patent attorneys handle litigation independently of a solicitor or barrister. A solicitor or barrister who specialises in patent law doesn’t have to be qualified as a patent attorney, and although it is technically possible to do so, few actually are.

[0] Technically, the UK classifies patent attorneys as legal professionals-in popular parlance, “lawyers”-but unlike solicitors and barristers, who are generalist legal professionals permitted (in principle) to practice in any legal area, a patent attorney is a specialist legal profession whose scope of legal practice is officially restricted to intellectual property (they are allowed to advise on non-patent IP law matters such as copyrights, and also related commercial law matters such as IP contracts, but not on legal matters which have no connection with IP)


Yes, it's quite jurisdiction dependant - I'm actually in Canada, not the US, and while my wife works with various foreign associates and I expect is familiar with that UK distinction, that's not a point I was clear on. :)

The patent is valid in several jurisdictions. It is easier to attack in the US due to the A and B combination.

The developer whom Synaptics is legally threatening in this case is in the UK.

Invalidating the US patent would do nothing to stop Synaptics from continuing to threaten that individual developer.

Invalidating the UK patent quite possibly would.


THis may be the case or may not be the case. I prefer to hit where it hurts. And it will hurt more in the US and I know US Patent law better and I see the opportunity in the US. If Synaptic wants desert, we can take care of the UK patent afterwards.

> My only concern is, what if some well-meaning person puts in a poorly constructed claim, and then that creates a counterproductive precedent.

If this were possible, patent holders would surely already be gaming it.


I think I may be able to hammer something out. Maybe this can be a group effort.

But this is not how you approach Synaptic. You write the counter and tell them they have 2 weeks to give you a free license or you file. Tick tock tick tock tick tock.


Probably not. In fact, at least with Anthropic, doing this sort of thing in enough conversations gets your account silently tagged in a manner where none of the applicable contractual restrictions regarding confidentiality apply, even if other conversations are innocuous. This is assuming ZDR is not enabled. You can see evidence of that in Anthropic's "Detecting and countering misuse of AI: September 2026" report. Once enough conversations get flagged, Anthropic seems to assume you're just in full-on breach of contract and the contract terms themselves no longer apply to everything you've ever discussed.

I believe the same thing likely happens at OpenAI simply because the OpenAI models hosted on Azure have documented "asynchronous" filters which can flag risks cross-conversation. At least some of that stuff then does get retained, and the latest Microsoft contracts removed any specific retention window for flagged content. Again, assuming MAM (Microsoft's "ZDR") is not enabled. This type of architecture at Azure probably didn't arise in a vacuum. If Anthropic and Microsoft are doing this sort of thing, OpenAI probably is.

Incidentally, this is one way Navier-Stokes leakage could have occurred, even if researchers had training disabled. Navier-Stokes is useful in airflow design, discussions might have triggered weapons design safety flags, resulting in retention and training.


> Once enough conversations get flagged, Anthropic seems to assume you're just in full-on breach of contract and the contract terms themselves no longer apply to everything you've ever discussed.

For which they should be dragged in front of a court and sued for millions and billions, if it is found out, that they were simply wrong. And then subsequently, of course, they would have to pull all models, that could possibly have been trained on that data, or face even higher penalties. Haha, but of course I am just dreaming here.


The FTC has issued a number of warnings to the LLM providers regarding lack of affirmative consent or shifting consent for user data handling, noting they could constitute unfair or deceptive trade practices, e.g.: https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/202...

There probably isn't any appetite at the FTC these days to actually follow through and investigate or possibly prosecute though. Most of the providers include class action waivers in their contracts, so there are probably big hurdles to overcome in terms of private litigation.


Even if this happens, it would still be worth it

Only way for penalties to sting is with prison time


> For which they should be dragged in front of a court and sued for millions and billions, if it is found out, that they were simply wrong. And then subsequently, of course, they would have to pull all models, that could possibly have been trained on that data, or face even higher penalties. Haha, but of course I am just dreaming here.

Nah, you got it backwards: the frontier labs don't answer to the courts. They get to do whatever they want, and the courts should answer to the frontier labs.


Incidentally by design, I'd bet. "Ooo Lean! Terrence Tao could be trying to break RSA or elliptic curve cryptography! /s" Conveniently for the PR team, we need this data to train on to make sure America has the biggest and best math.


Thank you for pointing this out. Dahl is not really an Akkadian-focused scholar, though he teaches it; he's more well-known for his work on proto-Elamite. Still, probably more than capable to assist. I wonder if any Akkadian experts have weighed in on this translation?


Cuis is the easiest entrypoint into modern Smalltalk. All the others have so, so much stuff (great stuff, just a lot of it). Cuis takes the approach of paring things down to something minimal. You can understand everything from top to bottom fairly easily.


It's got a very interesting pixel layout, different even from previous tandem OLEDs. Hard to describe quickly in words, but shown in Figure 2 here: https://www.apple.com/ca/macbook-pro/proled-xdr-display/pdf/... I imagine some of the C-shaped empty space is probably to accommodate the per-pixel emitter transistor.


This looks the same as the iPad Pro tandem OLED pixel layout. That I happen to write this comment on.


Doesn't the iPad Pro tandem OLED look like this? https://www.ltec-biz.com/en/report/post-3026/

The layout shown in that document is quite different.


A surprising amount of people use the back camera for document scanning.


Probably if the (in)famous airdrop was more reliable, people could figure out they can scan with their phone and airdrop to their tablet.

Too bad though, there's no way to put a scanned document onto iPad in Apple's ecosystem other than the otherwise useless camera.


Some people don't have iPhones though. Rare, I know. I have an iPad and my wife has an iPhone but I have never owned one at all and instead have always used Android right from the HTC G1.


The payphones aren't really all that anachronistic. Cue Phoebe Bridgers' 2020 song "Kyoto": "They still got payphones / It cost a dollar a minute".

That last point might change as you get deeper into the novel.

Still such a great novel; has aged very, very well.


Yup.

I read it just before Cyberpunk came out, just to know some more history of what I was stepping into. I was completely flabbergasted by how modern it felt to me.

I don't have the patience to read much anymore unless it is real good. I ate that one, more than once. And the others.


Payphones are also somewhat common in, off the top of my head, Canada and Australia. I'm sure plenty of other countries as well.

I wonder why the US went in so hard on removing or deactivating them (though we still have them here and there). Capitalism and not wanting to maintain stuff that's not used often, I guess.


Poor people used pay phones more often and once people with money could easily buy cell phones every time they saw some poor person use a payphone they would think, in large part thanks to hollywood, that they were drug dealers or otherwise up to no good. And once pay phones started becoming less common they became points for cops to harass poor people, because poor people are easy to get convictions on.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: