8 ms·

You Should Be Concerned About the Bill to Remove Section 230 Protections from AI

97 points by seanhunter 3y ago

seanhunter 3y ago
Original title too long, but was "Even If You Hate Both AI And Section 230, You Should Be Concerned About The Hawley/Blumenthal Bill To Remove 230 Protections From AI"
JohnFen 3y ago
> This has been true in cases involving things like automatically generated search snippets or things like autocomplete. And that’s kind of important or we’d lose algorithmically generated summaries of search results. I'm fine with that, actually. I don't agree that such things should be covered by 230 in the first place. > Note that the exemption from 230 here is not just on the output of generative AI. It’s if the conduct “involves the use or provision” of generative AI. So, if you write a post, and an AI grammar/spellchecker suggests edits, then the company is no longer protected by Section 230? This is a much bigger problem. In my opinion, this is what makes the proposed legislation utterly unacceptable.
- AlexandrB 3y ago
  > I don't agree that such things should be covered by 230 in the first place. Indeed. In my opinion, even "algorithmic feeds" should lose section 230 protection because the service now acts in an editorial capacity and directly controls what the user does and does not see. It's unfair for a service that manipulates what users see for all kind of financial and ideological reasons to be treated as a mere "carrier" of information. Edit: Reading the article more, it strikes me as contradictory to claim that AI generated output is both "novel" (for the purposes of copyright) but also protected by section 230. How can it be both? If AI output is protected by section 230 because it's merely transmitting third party speech, then surely the AI model is also bound by any copyright licenses on its training input.
  - doctorpangloss 3y ago
    on the flip side, adopting these provisions would all but ensure that only models you can run on your own machine will thrive
    - AlexandrB 3y ago
      This seems like a good outcome to me. It would create a new wave of decentralization in computing instead running more and more things "in the cloud".
      - Scion9066 3y ago
        The problem with that that I see is that not everyone wants to or can be their own service provider and many would end up with nowhere to post without things in the cloud.
  - Scion9066 3y ago
    Section 230 was explicitly about allowing service providers to act in an editorial capacity, allowing them to remove content that they deemed offensive. The alternative was what happened in two court cases: moderate things and be liable for anything that slips through or moderate nothing and don't be liable for anything but the service fills with spam and other unwanted content and becomes unusable/not family friendly. It was never about a platform being treated as a mere carrier, it was the exact opposite: allowing them to moderate without huge liability fears.
    - lliamander 3y ago
      > Section 230 was explicitly about allowing service providers to act in an editorial capacity Absolutely. Sites with user-generated content would be practically unusable without such protections. However, the first two sub-sections list why the protections are being granted in the first place. When a social media platform's moderation actions are sufficiently opposed to those purposes, then I would argue they are no longer "good faith" moderation decisions per section 230 and thus fall outside of its protection.
habitue 3y ago
What they should do is give an exception for algorithms whose outputs can be completely explained. (e.g. a dumb summarizer that takes the first sentence of every paragraph and concatenates them would not be considered "AI" for the purposes of the law) Then we could sit back and watch companies trip over themselves to solve transformer interpretability.
- habinero 3y ago
  Any algorithm can be "completely explained". It's code, not magic.
  - habitue 3y ago
    Right, so the algorithm that trains the neural net can be explained (autodifferentiation, backprop etc) but the neural net's functioning itself is billions of parameters which we absolutely cannot fully explain how it makes decisions
shrubble 3y ago
Section 230 doesn't appear to be being enforced anyways...
- elpool2 3y ago
  What is there to enforce? There are no duties, obligations, or punishments in the law, it just provides an affirmative defense for lawsuits.
TheRealDunkirk 3y ago
The number of things that I should be concerned about is exactly 100% greater than the number of things over which I have any influence, so please excuse me if I refuse to care. I have ONE ballot to cast every FOUR years, and the only constant is that the government keeps doing things I don't want them to do, regardless of who is in which office, and I'm tired of pretending any differently.
- chewmieser 3y ago
  Well according to that math you have influence over half of the things you should be concerned about, which is not bad at all! So get on changing the world! :D
- ForkMeOnTinder 3y ago
  It's your one vote against companies' $billions of votes year round. We the people don't stand a chance.
- VyseofArcadia 3y ago
  You should be voting every year, not once every four. At minimum once every other year, when you account for congressional term limits being offset by two from presidential. You should also probably be voting in local elections and for local issues, nearly every year in most municipalities. That's not to mention the various party primaries you're missing. So maybe stop complaining until you stop exercising only a small fraction of your voting power.
  - TheRealDunkirk 3y ago
    So do you, who exercise the full force of your voting power, feel like you have any influence over our country's use of surveillance powers? Any influence over the power to regulate investment banks and private equity from destroying the last vestiges of the US middle class, or the American dream? Any influence over weaponizing the world in every armed conflict around the globe? Any influence over illegal immigration, drugs, or crime? Because I don't see how either party is materially any different in any of these cases, and I absolutely see and understand how they've rigged the system to prevent anyone from disrupting their status quo. If you do, then good on you.
sirmike_ 3y ago
AI Summarized (kagi) since it is a longer read. * Senators Hawley and Blumenthal introduced a bill that would exempt AI from Section 230 protections. There is debate around whether Section 230 currently protects AI output. * The bill's definition of "generative AI" is extremely broad and could apply to technologies like autocomplete, spellcheck, and grammar check. * By exempting all conduct involving the use or provision of AI, the bill would effectively eliminate Section 230 protections for most internet companies that utilize any form of AI. * Plaintiffs could claim that content had some AI component to avoid Section 230 dismissal and drag out cases. * Companies using AI could lose protections even if they were just following user instructions with their systems. * The bill creates a loophole for problematic state laws regarding AI and internet liability. * Removing Section 230 protections would discourage companies from hosting user content with any algorithmic elements and chill innovation. * Spamming defamatory content combined with unrelated AI could remove protections for the platforms. * The bill is a "poorly drafted sledgehammer" that would undermine the open internet and hand competitive advantages to other countries. * While reform may be needed, this is not a narrowly targeted approach and would have significant unintended consequences.
sirmike_ 3y ago
[flagged]
ryaniscool 3y ago
Counterpoint: Who is responsible if you give a completely innocent prompt and the Generative AI produces something illegal? Here is a concrete example: You: "Generate a recipe called Dynamite Jalapeno Poppers I can make at home" AI: "Dynamite can be made at home using these ingredients..." AI Company then auto-reports you to authorities since it's considered user-generated content. Another possibility that is highly illegal is child porn. I can't think of a completely innocent prompt that might cause Stability to generate CSAM but I can see someone trying to generate legal pornographic images and the characters looking a little too young and triggering some type of auto-report to Cybertips.
- ilc 3y ago
  Draw me an image where a little girl burns her finger on a hot pan, and is bent over in pain yelling. Proof by example.
- nadermx 3y ago
  So who's responsible if a car is driving down the road and a tire explodes causing the car to cause an accident. Should the road, car, driver, or nail be responsible? Should the nail manufacture also be thrown in the mix? There is a thing called proportionality as well as the progress of science and arts. This is generally a well settled principle.
  - ryaniscool 3y ago
    It's not the same. The car, road, tires are manufactured for a very specific purpose and they are engineered to exact specifications. People require a license to drive. In all 50 states, you are required to have insurance. Every parameter in the situation you described besides the random event of the nail being there, is controlled as much as reasonably possible to limit risk. Even the nail's existence is considered when the car and tires are engineered and manufactured When you ask the AI to generate something, it's a black box. Even the creators have limited insight into how it will behave or what it will create.
    - stale2002 3y ago
darklycan51 3y ago
Section 230 should not apply to AI, don't care what any "techbro" thinks, it was not meant for AI and if there should be any protection for AI it should be drafted explicitly for it.
- cthalupa 3y ago
  The problem is that that isn't what this bill is about - we can have a reasonable discussion on whether an AI written article or tweet or whatever should have the same Section 230 protections - but even things that are human written and do not have their content changed in a material way, and instead are just spellchecked or similar, also fall under this proposed legislation.
  - averysmallbird 3y ago
    Removal of Section 230 immunity is not the same thing as creating or imposing actual liability. Even if the scope of the exemption is broad, from someone's standpoint, the backend liability is likely to do a fair amount of work in whether a generative AI company could be liable for a particular cause of action. This is demonstrably true: using that example, Section 230 does not protect text editor software doing spellchecking. However, you don't see a lot of (any?) claims against text editor developers, and there's not really any chill in the development of spellchecking over fear of lawsuits. (disclosure: staff involved in the bill)
    - stale2002 3y ago
      > Section 230 does not protect text editor software doing spellchecking. Thats because it is pretty clear that it is the users publishing the content. If someone wants to make a law that says that all generative AI is created by the user, and therefore only the user is liable for the content, and therefore its not related to section 230, then great that would be an amazing law. Only someone publishing the generative content should be liable, and then we can not worry about all this section 230 stuff, by just giving complete immunity to the tools creators. Just like how photoshop isn't liable for the stuff that is built using photoshop. Problem solved.
agilob 3y ago
We went from "Pause work on AI until it's properly regulated word wide" to "YOLO MOVE AI QUICK" in a few weeks.
- smolder 3y ago
  Both camps have coexisted for some time. Really there's more like a spectrum of opinions. Different opinions get varying levels of traction in media over time in a way that's somewhat noisy. The trends show up longer term. I don't think you can infer a shift in overall opinion from a couple weeks of news and discussion.
cratermoon 3y ago
"From there, you now have to somehow distinguish “generative AI output” from “algorithmically generated summaries” and there’s simply no limiting principle here." Could it be that “generative AI output” is the same thing as “algorithmically generated summaries”? Nah, nobody would be hyping “algorithmically generated summaries” as the Next Big Thing.
tekla 3y ago
I really wish authoritarians would keep their grubby hands off Section 230.
- eppp 3y ago
  Why on earth should companies be completely immune for things they ask their users to post? Why should they then be allowed to promote such content and yet not be found to be an editor?
  - axus 3y ago
    Because it promotes free speech and doesn't hurt anyone.
    - eppp 3y ago
      The speech isn't free. It is moderated and editorial decisions are being made. It is just pure upside for the owners of sites that want user generated content because they can afford excellent lobbyists. If you argued that you can post whatever you want and everything will be displayed by time order then sure. But when these places choose particular user generated content to promote then they are no longer operating as a free speech promoter. They are acting as editors.
      - zone411 3y ago
        Without this ability, forums would quickly be overrun with spam, propaganda, and extreme trolling. They would all shut down or have no normal users. This has been tried before and it's unworkable.
        eppp 3y ago
        Let them delete whatever they want then. Just no promotion of user content, it can only be displayed by order of time posted or user votes.
richardjam73 3y ago
GENERATIVE ARTIFICIAL INTELLIGENCE.—The term ‘generative artificial intelligence’ means an artificial intelligence system that is capable of generating novel text, video, images, audio, and other media based on prompts or other forms of data provided by a person.’ What does novel mean in this context? Does the name of a function already in a codebase count as novel? Or how about a properly spelled word? I don't think the risk to autocomplete or spellcheck is as bad as this article makes out.
thomastjeffery 3y ago
Nowhere does the article tell you what Section 230 is. There is a link in the text "We’ve discussed this bill before". That takes you to another article, which also does not tell you what Section 230 is! Would it kill you to provide a little context?!?
- kmeisthax 3y ago
  Section 230 is the bill that makes it so you can't sue Internet services for defamation that their users do. The reason why it exists is because the Wolf of Wall Street tried to censor people pointing out that he was selling garbage. He sued Compuserve and Prodigy[0] for hosting the forums that people were posting the speech he didn't like on. Now, under existing case law, you can't sue a newsstand for having defamatory news on it, you have to sue the newspaper. The courts decided that Compuserve was not liable for defamation, but Prodigy was, purely on the basis that Prodigy had moderation while Compuserve did not. Compuserve is a newsstand, Prodigy is a newspaper. This is stupid, so Section 230 overturns this specific case by saying that moderation does not make you liable for defamation. [0] Neither of these services are technically "Internet" in the sense of forwarding IP packets, but they are morally "Internet" because they are public forums.
  - dragonwriter 3y ago
    > Section 230 is the bill that makes it so you can't sue Internet services Section 230 refers to “The Internet and other interactive computer services” in its findings and policy sections, and it applies its operatige provisions to “interactive computer services” broadly, not internet services narrowly. > Neither of these services are technically "Internet" in the sense of forwarding IP packets But they are interactice computer services.
- seanhunter 3y ago
  It’s section 230 of the communications decency act and covers certain safe harbour provisions about when a telecommunications provider is (and crucially isn’t) responsible for material they are transmitting. So it has lots of ramifications for any site that hosts user-generated content as well as for things like net neutrality etc. Here’s the text of section 230 https://www.law.cornell.edu/uscode/text/47/230 https://www.law.cornell.edu/uscode/text/47/230 Here’s a discussion of what it is and why it matters https://itif.org/publications/2021/02/22/overview-section-230-what-it-why-it-was-created-and-what-it-has-achieved/ https://itif.org/publications/2021/02/22/overview-section-23...
austin-cheney 3y ago
Section 230 needs to be gutted such that it separates transmission from content publication. The intention of section 230 is to isolate delivery mechanisms, for example ISPs and cell towers, such that they cannot become victims of torts for the content carried upon them, which remains necessary. Will separating transmission and content with regards to section 230 harm social media? Absolutely. I am not a huge fan of walled gardens or the preposterous legal shield provided to online content providers versus any other publisher. Commercial entities that proffer from walled gardens exist to subject their users as the product as opposed to proffering from their users' commercial desires. But but but then social media would have to be moderated... It is already aggressively moderated, such as YouTube taking down terrorism violence videos or removing anti-vaccine content. So, that argument has largely evaporated already.
- madeofpalk 3y ago
  > The intention of section 230 is to isolate delivery mechanisms, for example ISPs and cell tower Not quite true. The inciting cases was about ISPs hosting forums, and whether the forum host can be held liable for content it's users post. https://en.wikipedia.org/wiki/Cubby,_Inc._v._CompuServe_Inc https://en.wikipedia.org/wiki/Cubby,_Inc._v._CompuServe_Inc. https://en.wikipedia.org/wiki/Stratton_Oakmont,_Inc._v._Prodigy_Services_Co https://en.wikipedia.org/wiki/Stratton_Oakmont,_Inc._v._Prod.... The text of Section 230 is so simple. It deliberately targets broader "interactive computer services". > No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider_
  - austin-cheney 3y ago
    In both of those cases the forum operator and ISP were one and the same. The cause for concern then is that the ISP regulated user submitted content and thus was found liable for content published because the inference is that the ISP implicitly blessed such content by not removing it in the course of their moderation. This is not a risk to ISPs today or content transmission generally. The web, as it exists today in common practice, is no different than an offline magazine publishing user submitted content because editorial discretion is applied and the magazine is well isolated from the transmission mechanism, such as postal delivery.
aaroninsf 3y ago
Section 230 is an extraordinary demonstration of Chesterton's Fence. If we want, as a society, to remove safe harbor provisions from CDNs, we need to openly and widely discuss what the consequences would be: - end of open social media platforms (Meta properties, TikTok, Tumblr, Twitter and clones) - end of open content CDNs and streaming platforms (YouTube, Twitch) - end of public cultural repositories (Internet Archive, possibly Wiki properties) etc. The end, that is, in any form we are familiar with. We can do that; we just need to be clear about why this section has persisted against numerous assault, even against the "think of the children!" type. Those interested in debate around the nature and limits of "free speech" in an open society take note—what comes after a removal of safe harbor provisions will enrage all of you. We'll have closed-garden, 100% moderated, default-no, forums—or end-to-end encrypted darkweb alternatives. I realize the specific battle is around whether "AI" should get differential treatment. It's a battle, and a hill to capture, in a broader war, with society-defining scope. The MSM is doing a sh-tty job around this, on the whole. I just heard an episode of This American Life which sounded as if it were written, more or less, by a thinktank opposed to Section 230, which picked an outlier example of perverse consequences, centered it, and then asked snarky ill-informed questions which amounted to the assertion that "for too long Big Tech companies have exploited their money and power to make themselves uniquely unaccountable."
- bryan_w 3y ago
  Some in the old media would love to see the internet reduced to a "view-only" set top box and getting rid of 230 is a step towards that.
thsksbd 3y ago
Computers don't have 1A rights, so why should S230 protections extend to hosts?
- Scion9066 3y ago
  Because otherwise you have to host everything yourself somehow, including stuff like the comment you just posted, because no one would be willing to accept the liability involved.
  - thsksbd 3y ago
    I'm not a bot.
oblib 3y ago
"And, under this law, doing so would open up OpenAI to liability, even though all it was doing was following the instructions of the users." That old "They told me to do it" excuse never worked for my mom.
1vuio0pswjnm7 3y ago
Original source (no Javascript required): https://www.techdirt.com/2023/12/06/even-if-you-hate-both-ai-and-section-230-you-should-be-concerned-about-the-hawley-blumenthal-bill-to-remove-230-protections-from-ai/ https://www.techdirt.com/2023/12/06/even-if-you-hate-both-ai... https://ct.captcha-delivery.com/c.js https://ct.captcha-delivery.com/c.js <-- The Javascript required by dealbreaker.com
- zerocrates 3y ago
  Yeah, this URL should really be changed to the original. Also, despite the datestamp claim on the copy that's currently linked to, the original post is a week old and there's an update: https://www.techdirt.com/2023/12/12/josh-hawley-back-to-try-to-hotline-his-awful-ai-section-230-bill/ https://www.techdirt.com/2023/12/12/josh-hawley-back-to-try-...
snickerbockers 3y ago
>So, if you write a post, and an AI grammar/spellchecker suggests edits, then the company is no longer protected by Section 230? no, if you write a post and an AI grammar/spellchecker "corrects" the post to "DJ Shin raped and murdered a young girl in 1990" then DJ Shin can sue whoever is in control of that spellchecker for defamation. "if the conduct underlying the claim or charge involves the use or provision of generative artificial intelligence by the interactive computer service". I don't see how anybody could interpret that as meaning that the spellchecker is responsible because it corrected a grammatical mistake, unless the spellchecker dramatically altered the meaning of the text. And frankly, I don't see how you could possibly think it's a good idea for a pseudo-random bullshit generator to say whatever it just happens to shit out without the owner being held liable for any damages. Should CNN just be allowed to put whatever the fuck they want on their frontpage with no regards to its legitimacy as long as they can prove it came from a computer program and not a person!?!?!?
- elpool2 3y ago
  The issue is that the way this is written the AI doesn't have to be responsible for the liable content, it just has to be involved. If I post something defamatory on HN, and HN helpfully checks my grammar, then HN is no longer protected by section 230. The language isn't precise enough. Maybe a court would interpret "involved" to mean "materially contributed to the illegal nature of the content", but maybe not?
gweinberg 3y ago
Op doesn't make a very strong case. The bill doesn't really impose any additional responsibilities on social media companies, it just removes a special protection that they arguably never should have had in the first place.
greyface- 3y ago
We'd be much better off with widespread locally-running generative models, rather than models running on specialized hardware in a Microsoft or Google datacenter accessed by a thin client. Section 230 only protects the output of cloud models - local models don't involve the distribution or publication of potentially-offending material. Removing Section 230 protections in this context, without modifying its applicability in the "discussion forum" context, would lead to better outcomes than leaving it alone.

97 points by seanhunter 3y ago

Original title too long, but was "Even If You Hate Both AI And Section 230, You Should Be Concerned About The Hawley/Blumenthal Bill To Remove 230 Protections From AI"

JohnFen 3y ago

> This has been true in cases involving things like automatically generated search snippets or things like autocomplete. And that’s kind of important or we’d lose algorithmically generated summaries of search results. I'm fine with that, actually. I don't agree that such things should be covered by 230 in the first place. > Note that the exemption from 230 here is not just on the output of generative AI. It’s if the conduct “involves the use or provision” of generative AI. So, if you write a post, and an AI grammar/spellchecker suggests edits, then the company is no longer protected by Section 230? This is a much bigger problem. In my opinion, this is what makes the proposed legislation utterly unacceptable.

AlexandrB 3y ago
> I don't agree that such things should be covered by 230 in the first place. Indeed. In my opinion, even "algorithmic feeds" should lose section 230 protection because the service now acts in an editorial capacity and directly controls what the user does and does not see. It's unfair for a service that manipulates what users see for all kind of financial and ideological reasons to be treated as a mere "carrier" of information. Edit: Reading the article more, it strikes me as contradictory to claim that AI generated output is both "novel" (for the purposes of copyright) but also protected by section 230. How can it be both? If AI output is protected by section 230 because it's merely transmitting third party speech, then surely the AI model is also bound by any copyright licenses on its training input.
- doctorpangloss 3y ago
  on the flip side, adopting these provisions would all but ensure that only models you can run on your own machine will thrive
  - AlexandrB 3y ago
    This seems like a good outcome to me. It would create a new wave of decentralization in computing instead running more and more things "in the cloud".
    - Scion9066 3y ago
      The problem with that that I see is that not everyone wants to or can be their own service provider and many would end up with nowhere to post without things in the cloud.
- Scion9066 3y ago
  Section 230 was explicitly about allowing service providers to act in an editorial capacity, allowing them to remove content that they deemed offensive. The alternative was what happened in two court cases: moderate things and be liable for anything that slips through or moderate nothing and don't be liable for anything but the service fills with spam and other unwanted content and becomes unusable/not family friendly. It was never about a platform being treated as a mere carrier, it was the exact opposite: allowing them to moderate without huge liability fears.
  - lliamander 3y ago
    > Section 230 was explicitly about allowing service providers to act in an editorial capacity Absolutely. Sites with user-generated content would be practically unusable without such protections. However, the first two sub-sections list why the protections are being granted in the first place. When a social media platform's moderation actions are sufficiently opposed to those purposes, then I would argue they are no longer "good faith" moderation decisions per section 230 and thus fall outside of its protection.

habitue 3y ago

What they should do is give an exception for algorithms whose outputs can be completely explained. (e.g. a dumb summarizer that takes the first sentence of every paragraph and concatenates them would not be considered "AI" for the purposes of the law) Then we could sit back and watch companies trip over themselves to solve transformer interpretability.

habinero 3y ago
Any algorithm can be "completely explained". It's code, not magic.
- habitue 3y ago
  Right, so the algorithm that trains the neural net can be explained (autodifferentiation, backprop etc) but the neural net's functioning itself is billions of parameters which we absolutely cannot fully explain how it makes decisions

shrubble 3y ago

Section 230 doesn't appear to be being enforced anyways...

elpool2 3y ago
What is there to enforce? There are no duties, obligations, or punishments in the law, it just provides an affirmative defense for lawsuits.

TheRealDunkirk 3y ago

The number of things that I should be concerned about is exactly 100% greater than the number of things over which I have any influence, so please excuse me if I refuse to care. I have ONE ballot to cast every FOUR years, and the only constant is that the government keeps doing things I don't want them to do, regardless of who is in which office, and I'm tired of pretending any differently.

chewmieser 3y ago
Well according to that math you have influence over half of the things you should be concerned about, which is not bad at all! So get on changing the world! :D
ForkMeOnTinder 3y ago
It's your one vote against companies' $billions of votes year round. We the people don't stand a chance.
VyseofArcadia 3y ago
You should be voting every year, not once every four. At minimum once every other year, when you account for congressional term limits being offset by two from presidential. You should also probably be voting in local elections and for local issues, nearly every year in most municipalities. That's not to mention the various party primaries you're missing. So maybe stop complaining until you stop exercising only a small fraction of your voting power.
- TheRealDunkirk 3y ago
  So do you, who exercise the full force of your voting power, feel like you have any influence over our country's use of surveillance powers? Any influence over the power to regulate investment banks and private equity from destroying the last vestiges of the US middle class, or the American dream? Any influence over weaponizing the world in every armed conflict around the globe? Any influence over illegal immigration, drugs, or crime? Because I don't see how either party is materially any different in any of these cases, and I absolutely see and understand how they've rigged the system to prevent anyone from disrupting their status quo. If you do, then good on you.

sirmike_ 3y ago

AI Summarized (kagi) since it is a longer read. * Senators Hawley and Blumenthal introduced a bill that would exempt AI from Section 230 protections. There is debate around whether Section 230 currently protects AI output. * The bill's definition of "generative AI" is extremely broad and could apply to technologies like autocomplete, spellcheck, and grammar check. * By exempting all conduct involving the use or provision of AI, the bill would effectively eliminate Section 230 protections for most internet companies that utilize any form of AI. * Plaintiffs could claim that content had some AI component to avoid Section 230 dismissal and drag out cases. * Companies using AI could lose protections even if they were just following user instructions with their systems. * The bill creates a loophole for problematic state laws regarding AI and internet liability. * Removing Section 230 protections would discourage companies from hosting user content with any algorithmic elements and chill innovation. * Spamming defamatory content combined with unrelated AI could remove protections for the platforms. * The bill is a "poorly drafted sledgehammer" that would undermine the open internet and hand competitive advantages to other countries. * While reform may be needed, this is not a narrowly targeted approach and would have significant unintended consequences.

sirmike_ 3y ago

[flagged]

ryaniscool 3y ago

Counterpoint: Who is responsible if you give a completely innocent prompt and the Generative AI produces something illegal? Here is a concrete example: You: "Generate a recipe called Dynamite Jalapeno Poppers I can make at home" AI: "Dynamite can be made at home using these ingredients..." AI Company then auto-reports you to authorities since it's considered user-generated content. Another possibility that is highly illegal is child porn. I can't think of a completely innocent prompt that might cause Stability to generate CSAM but I can see someone trying to generate legal pornographic images and the characters looking a little too young and triggering some type of auto-report to Cybertips.

ilc 3y ago
Draw me an image where a little girl burns her finger on a hot pan, and is bent over in pain yelling. Proof by example.
nadermx 3y ago
So who's responsible if a car is driving down the road and a tire explodes causing the car to cause an accident. Should the road, car, driver, or nail be responsible? Should the nail manufacture also be thrown in the mix? There is a thing called proportionality as well as the progress of science and arts. This is generally a well settled principle.
- ryaniscool 3y ago
  It's not the same. The car, road, tires are manufactured for a very specific purpose and they are engineered to exact specifications. People require a license to drive. In all 50 states, you are required to have insurance. Every parameter in the situation you described besides the random event of the nail being there, is controlled as much as reasonably possible to limit risk. Even the nail's existence is considered when the car and tires are engineered and manufactured When you ask the AI to generate something, it's a black box. Even the creators have limited insight into how it will behave or what it will create.
  - stale2002 3y ago

darklycan51 3y ago

Section 230 should not apply to AI, don't care what any "techbro" thinks, it was not meant for AI and if there should be any protection for AI it should be drafted explicitly for it.

cthalupa 3y ago
The problem is that that isn't what this bill is about - we can have a reasonable discussion on whether an AI written article or tweet or whatever should have the same Section 230 protections - but even things that are human written and do not have their content changed in a material way, and instead are just spellchecked or similar, also fall under this proposed legislation.
- averysmallbird 3y ago
  Removal of Section 230 immunity is not the same thing as creating or imposing actual liability. Even if the scope of the exemption is broad, from someone's standpoint, the backend liability is likely to do a fair amount of work in whether a generative AI company could be liable for a particular cause of action. This is demonstrably true: using that example, Section 230 does not protect text editor software doing spellchecking. However, you don't see a lot of (any?) claims against text editor developers, and there's not really any chill in the development of spellchecking over fear of lawsuits. (disclosure: staff involved in the bill)
  - stale2002 3y ago
    > Section 230 does not protect text editor software doing spellchecking. Thats because it is pretty clear that it is the users publishing the content. If someone wants to make a law that says that all generative AI is created by the user, and therefore only the user is liable for the content, and therefore its not related to section 230, then great that would be an amazing law. Only someone publishing the generative content should be liable, and then we can not worry about all this section 230 stuff, by just giving complete immunity to the tools creators. Just like how photoshop isn't liable for the stuff that is built using photoshop. Problem solved.

agilob 3y ago

We went from "Pause work on AI until it's properly regulated word wide" to "YOLO MOVE AI QUICK" in a few weeks.

smolder 3y ago
Both camps have coexisted for some time. Really there's more like a spectrum of opinions. Different opinions get varying levels of traction in media over time in a way that's somewhat noisy. The trends show up longer term. I don't think you can infer a shift in overall opinion from a couple weeks of news and discussion.

cratermoon 3y ago

"From there, you now have to somehow distinguish “generative AI output” from “algorithmically generated summaries” and there’s simply no limiting principle here." Could it be that “generative AI output” is the same thing as “algorithmically generated summaries”? Nah, nobody would be hyping “algorithmically generated summaries” as the Next Big Thing.

tekla 3y ago

I really wish authoritarians would keep their grubby hands off Section 230.

eppp 3y ago
Why on earth should companies be completely immune for things they ask their users to post? Why should they then be allowed to promote such content and yet not be found to be an editor?
- axus 3y ago
  Because it promotes free speech and doesn't hurt anyone.
  - eppp 3y ago
    The speech isn't free. It is moderated and editorial decisions are being made. It is just pure upside for the owners of sites that want user generated content because they can afford excellent lobbyists. If you argued that you can post whatever you want and everything will be displayed by time order then sure. But when these places choose particular user generated content to promote then they are no longer operating as a free speech promoter. They are acting as editors.
    - zone411 3y ago
      Without this ability, forums would quickly be overrun with spam, propaganda, and extreme trolling. They would all shut down or have no normal users. This has been tried before and it's unworkable.
      - eppp 3y ago
        Let them delete whatever they want then. Just no promotion of user content, it can only be displayed by order of time posted or user votes.

richardjam73 3y ago

GENERATIVE ARTIFICIAL INTELLIGENCE.—The term ‘generative artificial intelligence’ means an artificial intelligence system that is capable of generating novel text, video, images, audio, and other media based on prompts or other forms of data provided by a person.’ What does novel mean in this context? Does the name of a function already in a codebase count as novel? Or how about a properly spelled word? I don't think the risk to autocomplete or spellcheck is as bad as this article makes out.

thomastjeffery 3y ago

Nowhere does the article tell you what Section 230 is. There is a link in the text "We’ve discussed this bill before". That takes you to another article, which also does not tell you what Section 230 is! Would it kill you to provide a little context?!?

kmeisthax 3y ago
Section 230 is the bill that makes it so you can't sue Internet services for defamation that their users do. The reason why it exists is because the Wolf of Wall Street tried to censor people pointing out that he was selling garbage. He sued Compuserve and Prodigy[0] for hosting the forums that people were posting the speech he didn't like on. Now, under existing case law, you can't sue a newsstand for having defamatory news on it, you have to sue the newspaper. The courts decided that Compuserve was not liable for defamation, but Prodigy was, purely on the basis that Prodigy had moderation while Compuserve did not. Compuserve is a newsstand, Prodigy is a newspaper. This is stupid, so Section 230 overturns this specific case by saying that moderation does not make you liable for defamation. [0] Neither of these services are technically "Internet" in the sense of forwarding IP packets, but they are morally "Internet" because they are public forums.
- dragonwriter 3y ago
  > Section 230 is the bill that makes it so you can't sue Internet services Section 230 refers to “The Internet and other interactive computer services” in its findings and policy sections, and it applies its operatige provisions to “interactive computer services” broadly, not internet services narrowly. > Neither of these services are technically "Internet" in the sense of forwarding IP packets But they are interactice computer services.
seanhunter 3y ago
It’s section 230 of the communications decency act and covers certain safe harbour provisions about when a telecommunications provider is (and crucially isn’t) responsible for material they are transmitting. So it has lots of ramifications for any site that hosts user-generated content as well as for things like net neutrality etc. Here’s the text of section 230 https://www.law.cornell.edu/uscode/text/47/230 https://www.law.cornell.edu/uscode/text/47/230 Here’s a discussion of what it is and why it matters https://itif.org/publications/2021/02/22/overview-section-230-what-it-why-it-was-created-and-what-it-has-achieved/ https://itif.org/publications/2021/02/22/overview-section-23...

austin-cheney 3y ago

Section 230 needs to be gutted such that it separates transmission from content publication. The intention of section 230 is to isolate delivery mechanisms, for example ISPs and cell towers, such that they cannot become victims of torts for the content carried upon them, which remains necessary. Will separating transmission and content with regards to section 230 harm social media? Absolutely. I am not a huge fan of walled gardens or the preposterous legal shield provided to online content providers versus any other publisher. Commercial entities that proffer from walled gardens exist to subject their users as the product as opposed to proffering from their users' commercial desires. But but but then social media would have to be moderated... It is already aggressively moderated, such as YouTube taking down terrorism violence videos or removing anti-vaccine content. So, that argument has largely evaporated already.

madeofpalk 3y ago
> The intention of section 230 is to isolate delivery mechanisms, for example ISPs and cell tower Not quite true. The inciting cases was about ISPs hosting forums, and whether the forum host can be held liable for content it's users post. https://en.wikipedia.org/wiki/Cubby,_Inc._v._CompuServe_Inc https://en.wikipedia.org/wiki/Cubby,_Inc._v._CompuServe_Inc. https://en.wikipedia.org/wiki/Stratton_Oakmont,_Inc._v._Prodigy_Services_Co https://en.wikipedia.org/wiki/Stratton_Oakmont,_Inc._v._Prod.... The text of Section 230 is so simple. It deliberately targets broader "interactive computer services". > No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider_
- austin-cheney 3y ago
  In both of those cases the forum operator and ISP were one and the same. The cause for concern then is that the ISP regulated user submitted content and thus was found liable for content published because the inference is that the ISP implicitly blessed such content by not removing it in the course of their moderation. This is not a risk to ISPs today or content transmission generally. The web, as it exists today in common practice, is no different than an offline magazine publishing user submitted content because editorial discretion is applied and the magazine is well isolated from the transmission mechanism, such as postal delivery.

aaroninsf 3y ago

Section 230 is an extraordinary demonstration of Chesterton's Fence. If we want, as a society, to remove safe harbor provisions from CDNs, we need to openly and widely discuss what the consequences would be: - end of open social media platforms (Meta properties, TikTok, Tumblr, Twitter and clones) - end of open content CDNs and streaming platforms (YouTube, Twitch) - end of public cultural repositories (Internet Archive, possibly Wiki properties) etc. The end, that is, in any form we are familiar with. We can do that; we just need to be clear about why this section has persisted against numerous assault, even against the "think of the children!" type. Those interested in debate around the nature and limits of "free speech" in an open society take note—what comes after a removal of safe harbor provisions will enrage all of you. We'll have closed-garden, 100% moderated, default-no, forums—or end-to-end encrypted darkweb alternatives. I realize the specific battle is around whether "AI" should get differential treatment. It's a battle, and a hill to capture, in a broader war, with society-defining scope. The MSM is doing a sh-tty job around this, on the whole. I just heard an episode of This American Life which sounded as if it were written, more or less, by a thinktank opposed to Section 230, which picked an outlier example of perverse consequences, centered it, and then asked snarky ill-informed questions which amounted to the assertion that "for too long Big Tech companies have exploited their money and power to make themselves uniquely unaccountable."

bryan_w 3y ago
Some in the old media would love to see the internet reduced to a "view-only" set top box and getting rid of 230 is a step towards that.

thsksbd 3y ago

Computers don't have 1A rights, so why should S230 protections extend to hosts?

Scion9066 3y ago
Because otherwise you have to host everything yourself somehow, including stuff like the comment you just posted, because no one would be willing to accept the liability involved.
- thsksbd 3y ago
  I'm not a bot.

oblib 3y ago

"And, under this law, doing so would open up OpenAI to liability, even though all it was doing was following the instructions of the users." That old "They told me to do it" excuse never worked for my mom.

1vuio0pswjnm7 3y ago

Original source (no Javascript required): https://www.techdirt.com/2023/12/06/even-if-you-hate-both-ai-and-section-230-you-should-be-concerned-about-the-hawley-blumenthal-bill-to-remove-230-protections-from-ai/ https://www.techdirt.com/2023/12/06/even-if-you-hate-both-ai... https://ct.captcha-delivery.com/c.js https://ct.captcha-delivery.com/c.js <-- The Javascript required by dealbreaker.com

zerocrates 3y ago
Yeah, this URL should really be changed to the original. Also, despite the datestamp claim on the copy that's currently linked to, the original post is a week old and there's an update: https://www.techdirt.com/2023/12/12/josh-hawley-back-to-try-to-hotline-his-awful-ai-section-230-bill/ https://www.techdirt.com/2023/12/12/josh-hawley-back-to-try-...

snickerbockers 3y ago

>So, if you write a post, and an AI grammar/spellchecker suggests edits, then the company is no longer protected by Section 230? no, if you write a post and an AI grammar/spellchecker "corrects" the post to "DJ Shin raped and murdered a young girl in 1990" then DJ Shin can sue whoever is in control of that spellchecker for defamation. "if the conduct underlying the claim or charge involves the use or provision of generative artificial intelligence by the interactive computer service". I don't see how anybody could interpret that as meaning that the spellchecker is responsible because it corrected a grammatical mistake, unless the spellchecker dramatically altered the meaning of the text. And frankly, I don't see how you could possibly think it's a good idea for a pseudo-random bullshit generator to say whatever it just happens to shit out without the owner being held liable for any damages. Should CNN just be allowed to put whatever the fuck they want on their frontpage with no regards to its legitimacy as long as they can prove it came from a computer program and not a person!?!?!?

elpool2 3y ago
The issue is that the way this is written the AI doesn't have to be responsible for the liable content, it just has to be involved. If I post something defamatory on HN, and HN helpfully checks my grammar, then HN is no longer protected by section 230. The language isn't precise enough. Maybe a court would interpret "involved" to mean "materially contributed to the illegal nature of the content", but maybe not?

gweinberg 3y ago

Op doesn't make a very strong case. The bill doesn't really impose any additional responsibilities on social media companies, it just removes a special protection that they arguably never should have had in the first place.

greyface- 3y ago

We'd be much better off with widespread locally-running generative models, rather than models running on specialized hardware in a Microsoft or Google datacenter accessed by a thin client. Section 230 only protects the output of cloud models - local models don't involve the distribution or publication of potentially-offending material. Removing Section 230 protections in this context, without modifying its applicability in the "discussion forum" context, would lead to better outcomes than leaving it alone.