There are also stealth subscriptions.

Example: Google arbitrarily bricking Nest thermostats 1st and 2nd Gen to encourage purchase of Updated version (while the old devices still do go online in order to upload your data; they are artificially rendered useless). . That new hardware cost is a disguised subscription.

IoT hardware vendors have been doing this for quite a while -- often by discontinuing updates to Fix security defects their product was shipped with.
Or pushing out a deliberately customer-hostile update to lock features the product had been sold with.

Yuep. The TLDR for existing US case law is basically that anything publicly available on the open web

Yes and Google probably benefits from this precedent more than any other company.. They're running a frickin search engine. They hardly ever Ask or Get proper permission for scraping anything. Companies who publish News in particular could have massive claims against sites like google News for pulling their articles and displaying it on Google's site. Not only unauthorized scraping, but appropriation.

If this precedent gets overturned, then I can file a big Lawsuit against Google for scraping my web server without specific permission, And then move the case to turn the case into a $2 Trillion class action.

Opening of this is a treasure trove for the AI engines to feed on.

I doubt if the ACM will simply permit that one. They will likely have Open Access limited so have to register for a free open access account.
The verbiage implies you will be able to read ACM Journals, Proceedings, and Magazines.

It does not clearly state that you will be allowed full access to download PDF/text in an unlimited manner.

Meanwhile their website already has a discussion about Free vs Premium features. Also "Bulk citation export" and "PDF downloads" are listed as Premium features. I read that as they can still have any per-user usage restriction as they feel necessary that would also interfere with large-scale AI scraping.

See:

Basic features
Reading all ACM journals, proceedings and magazines
Search
Exporting citations

Sign in or upgrade to access Premium features, including:

Access to the ACM Guide to Computing Machinery
AI-generated article summaries
Podcast-style summaries of conference sessions
Advanced search
Rich article metadata, including download metrics, index terms and citations received
Bulk citation exports and PDF downloads

I don't see anything in the text of GPLv2 that says the seller is responsible for ensuring the buyer can exercise/fulfill those rights.

That would be a warranty. The seller opts to give a promise directly to the buyer that the buyer that they get the code
and can do the things that the GPL terms stated they are allowed to do.

The GPL does the opposite. It disclaims all warranties, unless the distributor provides you one in fact.
Including a copy of binaries based on GPL code does not automatically warrant that they follow the GPL.
Even giving the customer a copy of the GPL license does not provide a warranty of compliance or ability to comply with the GPL.

The GPL while disclaiming And not requiring any software devs to offer warranties; Also discusses situations where you could receive
GPL'd code, but not be allowed to distribute that code at all. That's kind of the opposite of a warranty: the text of the license describes situations where you can't distribute the software.

In any case your act of forking the code doesn't create a legal duty to you upon anyone else that didn't have one before.

Even if the upstream author's misconduct unknowingly causes your fork to become a copyright infringement; the GPL disclaims warranties.

There simply is not enough verbiage in the GPL to secure rights to the End Users against whatever wishes the authors of the original GPL software has in mind.
For example: If I write a program completely from scratch -- It is completely within my rights to publish that code as GPL, But also negotiate terms with various hardware vendors to my financial advantage, and provide a copy of the software with Alternate distribution terms which afford profits only to me.

That is one of the reasons; I believe it would make sense for the court to throw out the case; reaching the conclusion that Only the author of the Infringed work has any standing under GPL to enforce the license terms. What you the end user see as a "Violation" of your user rights - Is an exclusive right the author created that code still has a right to exploit. The GPL never forced them to dismiss that right and grant that right to the consumers of their code.

If the court says that the provisions of the GPL are invalid, that doesn't mean Vizio gets to just use the code

That is not on the table though.. It is more a question of What is the applicability and enforcement of the GPL, And who gets to enforce it, and what is the remedy. If it were just the authors of the software suing over GPL violation -- your remedy for breach of license would typically be money damages or Injunction - Pay us money; Plus cease and decist distributing the violative code. On the other hand "Release to us all additional source code you were obligated to have included as per the license agreement" would be an Equitable relief of specific performance -- extremely rare for courts to grant, unless it is part of some settlement agreement, and the judges would need the answer to the question - Why money and injunction alone are not enough to make the authors whole.

The GPL is a binding license agreement No doubt. What is extraordinary here is the party suing is not the copyright owner - they are suing on behalf of consumers. 3rd Party beneficiaries who are technically not parties to the license agreement.

The GPL is not between consumers and Vizio - it's between the authors of the software and Vizio.

So the courts could very well end up Dismissing the case making the determination that end consumers - the public in general does not get a remedy over the GPL violation, and only the authors of the software used are harmed, by having their copyright infringed. This is what you would expect. It will be an extraordinary earth-shattering thing if the court decides that the public and consumers in general have a remedy over GPL violations.

It makes sense, Because placing the code under the GPL is intended to be a gift to the public, so violating the terms is Unlawfully stealing not just from the author, but from the public ecosystem as well. On the other hand, courts have never acknowledge that as such with open source licenses before, And it is very possible they will end up not acknowledging it today.

But if the precedent is set that the public can sue over GPL violations, AND specific performance will be ordered (You must comply with the GPL by immediately distributing all witheld code. It is not that you can pay a few dollars to each class action member to settle the claims of a class.), then the GPL becomes much more powerful.

It seems to me that an AI running a vending machine, should be set up to only accept a limit selection of prompts.

Not really. That doesn't work, because the AI is tasked with managing the whole vending machine business.
It is not a simple product ordering AI. This is an AI that manages the business.
It does things like negotiate pricing; decide what types of goods will be stocked; decide which company things will be ordered to stock the machine in the first place. Contract negotiations usually involve fairly detailed email exchanges, and it would be easy for a media reporter to simply pose as a vendor and
inject some shlop into its conversations.

The role of a business manager includes answering customer inquiries, which could be anything, But that also includes managing bank accounts, sourcing products - writing and exchanging emails back and forth with vendors, etc. Trying to lock them down to some type of predefined prompt list would defeat the point that the purpose of the AI is to manage the whole vending machine business without human oversight.

The language model is "prompted" for that goal and the language model itself was not trained with that goal embedded into the training process.

The AI does not seem to have been programmed with the basic goal of making a profit.

The prompt probably started with that. The problem is ALL the data a LLM disseminates gets appended to the prompt. That is how an LLM works. Therefore.. by sending over new data you can manipulate the outcome.

For a Linux shell analogy.. the Initial system programming is like a .bashrc, And when your AI talks to people - they get access to a bash prompt. Of course they can manupulate the shell to override directives that came earlier from the preloaded .bashrc.

For a car analogy: You were taught to always come to a complete stop in driver's Ed. But on the road you have another driver shouting at you not to stop at the stop sign; So you get taught to cease stopping at stop signs.

Language models tend to give priorityTowards words that come later over words that come later. Also; your AI inference has a limited working memory size or context window.. Eventually you run out of space, And the earlier words need to be summarized in order to fit the entire combined prompt within the limited time and space for the AI.

If you Want to somehow prevent this; I am afraid the only answer is You will need to combine multiple AIs, and have a true supervisory process.
You need a dedicated AI to read the untrusted user input and Vet that input, and ensure that input is safe before any of it can be h anded over to the empowered AIs for processing. You need extra guardrails and filtering systems to make sure a clever prompt hacker does not persuade the supervisory AI to allow them a sandbox escape.

The Supervisory AI over the empowered AI, and the filtering AIs for user and untrusted inputs need to have their own Independent prompt stacks. The various filtering AIs and supervisory AIs need a number of specialized "panic buttons" that harness True stop controls instead of merely being able to provide suggestions or advise. For example: A supervisory panic initiated a predefined process and blocks further decisions from the impacted unit from being approved until a series of conditions are met to allow a status reset.

It seems like there needs to be a lawsuit initiated against Apple and the retailer for the full value of that account.

All I know about Volkswagon is they're a car company who deliberately cheated on their emissions tests.

No surprise their demand is falling away due to past outright illegal conduct.

Their license to manufacture a single new unit should have been cancelled the day this was found out. So I don't feel sorry for them.. surprised They did not have a complete shutdown imposed by the government sooner.

EULAs can not legally apply to a minor, just have your neighbor's kid set it up.

If you knew about the EULA, then hazarding to allow the minor to click accept still counts as you accepting it.

Judges are not keen to entertain "workarounds" like the kind you are describing. You can't avoid being deemed to have accepted a EULA by deliberately causing it to be accepted, no matter what method you pick.

I mean if such antics would work in practice; people could just let a cat click randomly; hex edit or NOP out the dialog display function from the executable, or figure out which bit written to disk or flash will bypass the prompt; etc. The concept of a clickwrap license would become a joke.

they sold it, as most people would to erase their streaming service credentials.

Your appraisal of consumer security awareness is way too optimistic.

Most people would just sell the TV. It's uncommon to seem them also reset to default aside from tested units sold by some secondhand stores that clean up used gear before selling.

Streaming services; assuming an old TV was even used for those; normally detect if a device has gone unused or moves to a different ISP or geolocation and cancel the device token requiring a revalidation. The old TV's reason for being for sale might even be that Netflix, etc, revoked their compatibility with it due to its age or outdatedness.

Anyway. There is never any presumption the seller factory defaulted their TV before selling it.

The manufacturer can log the EULA acceptance, and they will most likely be able to report on exactly the date, time, and IP address when someone clicked Okay. In the case of a dispute; the onus would be on the manufacturer to show evidence that the customer agreed. That is if the customer disputes the alleged fact that a EULA was accepted by them.

Get a new TV and never, EVER let it connect to the network.

Be really really careful. Manufacturers keep coming up with more and more ways to get it just enough internet access to talk to home even if you don't want it too. Hidden cellular modems. New mesh networking protocols like Sidwalk. Bluetooth. Aggressive wifi autoconfig. Ethernet over HDMI. etc

I am pretty sure they all spy on you now if connected to the internet. Some of them may be more obvious about it. Ideally you would get a TV that does not connect to the internet at all, Or turn the feature off, but A. It is almost impossible now, and B. Manufacturers are shady even if you turn the feature off.

You don't configure the WiFi: they will go into a loop searching for any AP they can connect to and grab any internet connection they can get to phone home, if possible; they'll especially try connecting to any unsecured AP that comes by.

It's kind of ridiculous the lengths they will go to; would probably make a good XKCD.

Possibly consider installing WiFi signal-blocking barriers around your TV, and force it to remain connected to a distinct SSID where it gets an IP from DHCP but has zero access to internet and zero access to LAN and other devices.