US GUARD Act: Age Verification for AI Chatbots
Posted by Hefty_Wolverine_553@reddit | LocalLLaMA | View on Reddit | 21 comments
There's been a growing number of AI regulation proposals I've been seeing in the US, and this bill in particular came to my attention today after seeing this article. The bill (which has just been "unanimously advanced to the Senate floor"), similar to other age verification policies, uses children's safety as a disguise to implement age verification for AI chatbots.
To require artificial intelligence chatbots to implement age verification measures and make certain disclosures, and for other purposes.
The wording of this bill is rather worrying (like many other invasive policies), and unfortunately I believe it may have a good chance of passing, with the US eagerly taking notes from the EU at the moment. As time goes on, and governments continue to restrict AI models and invade upon our privacy, I think more and more people will see the value in a local AI setup. I just hope that the current influx of open weights models will continue...
a_beautiful_rhind@reddit
It's not "age" verificiation. it's ID verification and forgive me if I don't believe they don't hold onto or link the information.
Sites like persona take biometrics of your face and then your ID. Hilariously they say to cover up any "private" information while simultaneously asking for the barcode on the back of the card and then failing you for hiding it. If they lie about something like this, they probably lie about everything.
Chatbots and "save the children" are a foot in the door to expand this everywhere. Pretty much you'll never have an account to anything again without submitting to 1984. Leaders in the EU said the quiet part out loud recently. "When someone says something online, I want to know who is saying it"
Doesn't matter that you live in the US, the same international companies are lobbying for it. Utah wants to hold sites liable for people using VPNs to bypass "age" verification. Sites will basically lock you out more than they do already.
"well I use local models" isn't going to help as they login and ID gate all the downloads. Github is owned by microsoft who are huge proponents. Even the stupid inference code is getting gated. The internet of the last 5-10 yeas is already less free and this is the plan for the next decade. Attacking unpopular things like "corn" or "AI" is how they manufacture consent to get the legislation on the books.
Far_Composer_5714@reddit
So it's not that they lie it's that they don't tell the truth. The person you submit the documents will tell you that they do provide the information to a third party.
So they will process it sell at to a third party and delete it.
That third party still has it but they deleted it...
a_beautiful_rhind@reddit
I can see that being the case too.
thread-e-printing@reddit
The Child is the emblem of futurity's unquestioned value. The "Children" they are interested in "saving" are only insofar as they are blank canvases on which to impress their images of family and nation.
sine120@reddit
It was never about saving the kids, and the trajectory is everything where you have an opinion/ get information. The whole industrial world is manufacturing a surveillance state.
ttkciar@reddit
On one hand, this isn't good.
On the other hand, bad bills get proposed all the time, and most of them never pass into law.
I wouldn't worry about this too much, but it's worth keeping an eye on, and everyone should be cloning repos and hoarding Huggingface models/data.
hejj@reddit
This would also no doubt benefit local AI, so even shitty laws can have a silver lining
therealpygon@reddit
Clearly someone didn't read the bill. If you had, you would not that there is NO limitations or carve outs for local AI.
(2) ARTIFICIAL INTELLIGENCE CHATBOT.—The term “artificial intelligence chatbot”—
(A) means any interactive computer service or software application that— (i) produces new expressive content or responses not fully predetermined by the developer or operator of the service or application; and (ii) accepts open-ended natural-language or multimodal user input and produces adaptive or context-responsive output; and (B) does not include an interactive computer service or software application— (i) the responses of which are limited to contextualized replies; and (ii) that is unable to respond on a range of topics outside of a narrow specified purpose.
(3) COVERED ENTITY.—The term “covered entity” means any person who owns, operates, or otherwise makes available an artificial intelligence chatbot to individuals in the United States."
Read it again.
"Any computers service or software that" "produces [...] new [...] responses" "to user input". Unless you've know of new definitions that I don't, that absolutely describes the software and models for local AI.
Are the people who wrote OpenWebUI required to obtain age verification before allowing you to download software that provides local AI responses? By this definition, certainly sounds like yes.
Is Github required to obtain age verification before allowing you to download said software? By this definition, also looks like yes.
Is Huggingface required to obtain age verification before allowing you download AI models that accept human language and provided non-predetermined responses? Again, certainly appears so.
Now we arrive at enforcement: What if huggingface and github, and all the chinese model providers do not to perform age verification? Could injunctions be sought to block ALL distribution until there is "compliance"?
Yes.
hejj@reddit
Much like the other age verification bills that are passing all over the place and the other use cases impacted by them, there's a certain demographic (particularly in the open source world) that doesn't intend to comply. I myself fall into that camp.
thread-e-printing@reddit
That's only an engineering problem, though
ComplexType568@reddit
Nice thing about local is that they will probably not care to even search nor actually enforce said laws
RedParaglider@reddit
If you buy your kid a gaming PC they can do whatever the fuck they want at varying speeds with AI. I'm glad nobody in the government is paying any attention to local systems.
Miriel_z@reddit
Hold on, so "manipulate emotions and influence behavior in ways that exploit the developmental vulnerabilities" of adults is perfectly OK? Like propaganda and shoving ads in my face?
thread-e-printing@reddit
In this country that's the exclusive right of good Christian parents
ortegaalfredo@reddit
I would say it's stupid but after seeing that they had to kill chatgpt-4o because people were falling in-love with it, maybe its not a such a bad idea.
my_name_isnt_clever@reddit
You're right, let's also dull every knife on the planet and make knife sharpeners scan your ID to protect the children.
Lesser-than@reddit
As always anything done in the name of children is rarely actually targeted at helping children. I do think we need to do something more about usage in schools. I cant imaging growing up in school right now with the answer to just about any assignment is one prompt away, I don't think I would have it in me to resist.
mystery_biscotti@reddit
In college, I used LLMs to explain to me coding concepts I was struggling with. I wasn't the kind of ND student that tutors enjoyed helping because the way I approach a topic can be a bit sideways from what they're used to. And there's a certain unspoken rule about the number of questions you're allowed to ask.
the-username-is-here@reddit
Think about the children!
Universal excuse for sleazebag politicians trying to build totalitarian state.
DataGOGO@reddit
Nope... they can fuck off.
NNN_Throwaway2@reddit
I'd say the wide support for age verification is shocking, but as people seem to be getting dumber by the day I guess it isn't that surprising.
Of course, the big justification for age verification is protecting children; yet at the same time, data brokering, (which is the sole main driver of the exploitation of children by online platforms) undergoes zero regulation whatsoever in the US and very little abroad. Age verification and real ID only plays further into mass data collection.
The implication? Age verification gets attached to "spooky" online services, while ostensibly kid-friendly but in reality highly manipulative content that is supposedly safe for children has no restrictions placed upon it, allowing future generations to be exploited and controlled by power brokers, which in turns allows them to influence future government policy.
In other words, safety only matters when it threatens public perception and the bottom line of tech companies.