jwitthuhn
jwitthuhn on Hacker News
It is a great example to refute the common misconception that LLMs are inherently non-deterministic. I see people repeating this a lot and being able to point at a CPU implemented within one is a very strong counter-example.
Then that piracy would be unethical behavior by the individuals doing it, not the one who released the jailbreak.
The person who released this is no more responsible for switch piracy than linus torvalds is responsible for me playing pirated NES roms on my linux laptop.
Failing to publish it is unethical because this information allows Nintendo's customers to use the hardware they purchased as they see fit.
They didn't release the patch out of fear that someone would quickly use the patch to develop an exploit, meaning an exploit is available shortly after the patch.
What they have achieved by delaying the patch is that exploits are now available publicly before a patch.
Surely this is the worse outcome.
Hopefully the industry will learn something about how withholding fixes to known vulnerabilities is, in fact, bad policy.
Seems like some of the people engaging in "responsible disclosure" were not very responsible and it leaked out.
With that info public they do really need to just release the fix, assuming that tweet is accurate.
edit: Mitigation for impacted versions is to set
`session.upload_progress.enabled = Off`
in php.ini
As implemented it leaks information to HIBP. Whether or not that is a risk depends on your threat model. Leaking information about your local environment is a fact regardless of your threat model.
This is standard tp-link policy. They release a model, give it almost no support, then release a new revision a year later with incompatible hardware and then the old one is never updated again.
So the solution to my iDevice dying is that I have to buy a new one to maintain access to my accounts and have no option to move to non-Apple hardware?
Surely it is possible to use AI and also comprehend its output.
Full disk access requires explicit consent from the user, and can only be given to an application using an obtuse workflow that is deliberately more complicated than just clicking accept.
Apple considers a user deliberately delegating access to their data to any program not controlled by Apple to be a serious problem that must be corrected.
Application developers are not making the user experience worse, Apple is.
Fortunately the corporations here lack the power to imprison people for saying things they don't like.
Apple also forces users onto a web browser that effectively has the same limitations as recent versions of chrome with respect to ad blocking.
They could easily resolve that by allowing alternate browsers, but choose not to.
That is a quote from the lead developer of WoW about how everyone asking for classic wow thought they wanted it but wouldn't actually play it. Then blizzard made classic wow and it was a huge hit.
The entire point is that it can't be reasonably enforced with any granularity. Rights-holders want it to be like that so a copyright win in a single country means something has to be taken down globally.
Dario's idea of an ethical focus seems to be keeping powerful models out of the hand of anyone unethical, which coincidentally is everyone except him.
Generative AI like LLMs and image diffusion models are perfectly capable of being deterministic. I have seen it repeated a lot that they aren't and that simply isn't true.
edit: to answer the second part, photoshop brushes are absolutely generative but I'm not sure they are AI.
I think people don't know what they mean and I have yet to receive a coherent response when asking such people what they mean.
As best as I can tell "Generative AI" is a common shorthand for "AI I don't like"
I'm quite curious about that claim that pitch correction is not generative AI. It literally generates a complete output waveform.
It is less secure though, and they are doing it in the name of security. People are right to call that out as bullshit.
Availability is a key component of security, and this is a 24-hour DOS on apps not blessed by google.
I have plenty of epithets for people who accuse others of lying based on vibes alone. It is not necessary or justifiable.
If models that fail the safety testing are banned, then asking for that testing is asking for a ban. That you believe any sane regulation would involve such a ban doesn't stop it from being a ban.
Surprised to see I partially agree. They are indeed ungovernable and that is the best part.
Not really, on the project I'm working on right now most of my longer prompts seems to consume somewhere between 300k and 400k tokens (input + output) and I find that both Opus 4.8 and Fable 5 don't seem to get lost with that much context. At least the results still look good to me.
I don't think I've ever pushed past 500k though, it may well fall apart at the higher end of what is officially supported.
Claude does the same when you hit the context limit, but the difference is that claude's context limit is 1m and Codex's is 272k now. Claude lets you get a lot more done before you have to worry about the start getting compacted away.
The text you linked mandates, as the first technical requirement, "An ADDW system shall determine when the driver’s visual attention is not directed towards the driving tasks and alert the driver through the vehicle human–machine interface."
Can you describe how you believe the driver's visual attention can be tracked by anything other than a camera pointed at the driver's face?
If no such other system exists, how is this doing anything other than mandating cameras pointed at the driver?
Why would physical copies matter for this? All physical games have to be signed by Sony anyways so it's not like a third-party can produce them.
This is true but the buyout happened in 1985. Servco Pacific has been in charge for longer than Leo Fender and CBS, the two previous owners, put together.