Hacker Newsnew | past | comments | ask | show | jobs | submit | aaronmdjones's commentslogin

> > AI training might be copyright infringement. But there’s no cases or laws to establish that.

> In September 2025, Anthropic agreed to pay at least $1.5 billion to settle a class-action lawsuit over using roughly 500,000 copyrighted books from "shadow libraries" to train their Claude LLMs.

Yes, but not because they were training LLMs with it. The judge in the case found specifically that training the LLMs on the copyrighted material was not copyright infringement; the only copyright infringement Anthropic had committed was acquiring the material itself. In other words, if they had legally bought all of the books they used, they would have been able to train their LLMs on them with no recourse from rights holders.


I also drew parallels to the Uberlingen mid-aid collision, but for a different reason.

The mid-air collision occurred because the Russian air crew maneuvered contrary to their TCAS instruction (it commanded them to climb, the controller ordered them to descend). They were not trained that TCAS is the ultimate authority in this situation; it exists precisely because the controller has already failed in their separation duties, and if you have TCAS giving you a resolution advisory, your aircraft is no longer under ATC control and you must ignore any ATC instruction to the contrary. The other aircraft was correctly following its TCAS instruction (descending) because their crew was trained in this. Both planes descended and still hit each other.

In this case, KLGA has RWSLs (Runway Status Lights), including RELs (Runway Entrance Lights) on taxiways, that behave like traffic lights on roads. This too is completely automated and is the last-ditch resort for when a controller has already failed in their separation duties. This system processes transponder data of nearby aircraft and determines whether an aircraft is about to take off (is on the runway and accelerating) or land (is approaching the runway and descending). In either case the RELs go red automatically, and the controller cannot override this.

The driver of the ARFF probably [1] placed more emphasis on the controller's clearance to cross than the lights telling him to stay put, in exactly the same way that the Russian air crew placed more emphasis on the controller's instruction to descend than their TCAS instruction to climb, not realising that they were maneuvering contrary to the thing that exists specifically to prevent these accidents.

EDIT: I am not assigning blame to the controller here. They are human, and humans make mistakes. That's why these systems exist. Having one person handle an airport the size of KLGA is an accident waiting to happen.

[1] Obviously this is unknown at this point, and is something the NTSB will investigate. The system could have been down for maintenance for example.


When I went to university 17 years ago, all of the computers (except the Macs) had dual-boot Windows 7 and Ubuntu 9.04.

I'll give you five guesses which OS I never booted into.


That's a bit of a trick question, because if you'd booted into Windows, it would have eventually broken the dual-boot.

I have a 16 GiB Optane NVMe M.2 drive in my router as a boot drive, running OpenWRT.

It's so incredibly fast and responsive that the LuCI interface completely loads the moment I hit enter on the login form.


> Work? Online.

Presumably my place of employment would have already verified my identity when I started working for them.

> School? Online.

Ditto, and for kids enrolling during lockdown they wouldn't have any ID to hand over anyway.

> Recreational activities? Online.

> Talking to loved ones you don’t live with? Online.

> Birthday party? Online.

> Nonfood shopping? Online.

Doesn't need me to verify my identity.

> Banking?

Every bank I've ever interacted with has done this verification in-person.

> Paying taxes

The government already knows who I am and what I look like (by issuing the ID to me); this is fine.

> and bills?

Direct debit just has a sort code, account number, and name on the account. No verification of identity.

> Job interview?

I wouldn't think that a new employer would be verifying your identity until you actually get the job offer?

> Doctors appointment?

Doesn't need me to verify my identity to arrange it. I have to go there in person anyway, I can damn well show them my ID then.

> Dating? You guessed it, online.

Eh. This one is a grey area. I can see the desire to have members verify who they are. I've also seen how badly that can go wrong.


It's better with absolutely no cooling. It doesn't even consume (and thus dissipate) 100mW flat-out.


Maybe they should have branded it differently …


They did, it's the Raspberry Pi Pico (as opposed to the Raspberry Pi) as a dev board or the RP2350 (as opposed to the BCMXXXX) as a chip.


Seriously. Different. Eh.


Generally speaking, there are two levels of crime in the US; misdemeanors and felonies. Both will land you with a criminal record, but a misdemeanor-only record will not show up on some standard background checks and does not remove your right to bear arms or vote, for example. Felonies are much more serious, and generally mandate a minimum prison sentence of 1 year unless plead down, while the sentencing for misdemeanors generally caps out at a year and typically just gets reduced to fines and community service, or a short stint (e.g. a couple weeks) in the local jail instead of a prison.

In some states, first offense non-violent felony convictions (e.g. exceeding the speed limit while fleeing police in a vehicle) can be expunged from your record when you turn 21 (if you were convicted and served out your sentence before turning 21). Otherwise felonies generally stay with you for life.


It also has various options to adjust the behaviour, from no blocks at all, to not even being able to charge the phone (or use the phone to charge something else) -- even when unlocked. Changing the mode of operation requires the device PIN, just as changing the device PIN does.

Note that it behaves subtly differently to how you described in case it was connected to something before being locked. In that case data access will remain -- even though the phone is now locked -- until the device is disconnected.


The author may just be showing their age a bit. That's what Ctrl+Alt+Del does on modern versions of Windows, but from Windows 95 to Windows XP (inclusive) it directly launched the Task Manager.


Ctrl+Alt+Del on an IBM PC or a compatible clone reboots the machine no questions asked. There's a dedicated reset button in case that fails.

Doing anything other than a reboot started with protected mode MS-Windows 3.1 IIRC (then marketed as "386 enhanced mode").


Yeah; in Windows 3.1, Ctrl+Alt+Del took you to a blue screen that allowed you to kill an unresponsive task (but didn't display a list of tasks; the Task List was launched with Ctrl+Esc), or told you there was no such task to kill if there wasn't.

Before Windows 3.1 it just rebooted the machine as you described.

Launching Task Manager was the 95 to XP behaviour, but NT behaved differently -- even Windows NT 4.0 (developed alongside Windows 95) took you to the security screen with Ctrl+Alt+Del (something that would later be ported to Vista), where launching Task Manager was one of its options. These OSes weren't used residentially though, until Windows 2000 attempted to merge their lineages and Windows XP finally cemented the deal.


I still have my Casio FX-83MS which my mother purchased for me when I started secondary school 24 years ago. It still works, it usually sits on my desk, and I use it a couple of times per month.


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

Search: