As I have pointed out before, rationalists & technology workers have a self-serving bias that makes them preferential to stories about public servants being stupid. This tweet about the NY AI bill is a great example of that:
The above post was quoted uncritically by Zvi, Eliezer, and Dean W. Ball, who each wrote some variation of “woe that state legislatures are so horrible”. The real bill is quite abit narrower than the tweet suggests; from GPT-5.4:
New York Senate bill S7263 is officially titled “Imposes liability for damages caused by a chatbot impersonating certain licensed professionals.” Its operative language says a chatbot proprietor may not let the chatbot provide a substantive response, information, advice, or take an action that, if done by a natural person, would constitute unauthorized practice or unauthorized use of a professional title under specified provisions of the Education Law or Judiciary Law.
That is materially narrower than a blanket ban on AI “answering questions related to” medicine, law, dentistry, nursing, psychology, social work, engineering, etc. The bill targets impersonation/unauthorized practice, not every answer touching those subjects. An official Senate press release describing the same bill likewise says it would prohibit chatbots from giving substantive responses “that can be mistaken for professional counseling” and identifies it as legislation to prevent AI from “impersonating certain licensed professionals.”
So the post overstates the bill’s scope: the bill does not ban AI from answering all questions related to those professions; it bars a narrower category of chatbot conduct that would amount to unauthorized professional practice if done by a human.
Public policymakers are ultimately the people who are going to be drafting laws & organizing responses about AI safety, and the reception of those laws is going to be driven partly by priors about their efficacy, developed from stuff like this. Boosting hyperbolic misinfo about particular bills, if it’s an attempt to build solidarity/shared intuitions with people in tech circles (which, to be clear, I don’t think this is), seems extremely counterproductive.
Sorry, I don’t understand, this seems like a straightforward ban:
Its operative language says a chatbot proprietor may not let the chatbot provide a substantive response, information, advice, or take an action that, if done by a natural person, would constitute unauthorized practice or unauthorized use of a professional title under specified provisions of the Education Law or Judiciary Law.
My understanding is that giving “substantive medical advice” is “unauthorized practice” and as such is banned. That’s what the tweet above says, pretty straightforwardly. The tweet is pretty clear that it is only including “substantive responses”, so I don’t think anyone is being misled.
I would be in favor of it saying “effective ban”, but honestly, it’s so close to just a straightforward direct ban that I think the current thing is totally fine. To be clear, the bill is directly banning AI systems giving substantive medical advice. The bill is quite clear that a disclaimer would not be sufficient, and AI systems would only be allowed to discuss general health considerations. Any kind of personalized medical advice would be banned, no matter what disclaimers you use.
On disclaimers, the bill literally says:
proprietor may not waive or disclaim this liability merely by notifying consumers that they are interacting with a non-human chatbot system.
My guess is if you look at the politicians who have been proposing this, they will also refer to this as “a ban on chatbots giving medical advice”. I haven’t looked this up, but I am a bit above 50% that the supporting side also thinks about this as a ban.
My understanding is that giving “substantive medical advice” is “unauthorized practice” and as such is banned.
It depends on what exactly you mean by “substantive medical advice” but you have to read these statutes against the backdrop of robust free speech protections when applied to giving advice.
This is a somewhat unsettled legal area and the Supreme Court is currently considering a case (Chiles v. Salazar) that may reshape the doctrine, but if you’re talking about professional practice that purely involves speech then it gets a lot of protection. That quickly recedes once any kind of conduct is involved (i.e., any kind of physical interaction), so the government usually wins unauthorized practice cases by pointing to something other than talking that was done.
Presumptively, if you’re only talking the government can’t regulate the content of that speech without clearing an exceptionally high legal bar and so someone giving you purely verbal medical advice without any physical component is likely in the clear. That’s why it’s not unauthorized practice of medicine if you tell me you have a headache and I suggest you take an aspirin, and all of the various websites out there that let you submit a list of symptoms and then suggest a diagnosis are fine too.
To crack down on something that doesn’t involve fraud/impersonation, the government would have to rely on the so-called “professional speech” exception to the first amendment, used in some circuit courts. It’s not clear this is even good law; the Supreme Court’s majority has never endorsed it, and they cast significant doubt on in in NIFLA v. Becerra (2018).
To the extent the professional speech exception exists at all, it applies when there is a “personal nexus” between the professional and the client. The threshold inquiry is whether or not “a speaker … purport[s] to be exercising judgment on behalf of any particular individual with whose circumstances he is directly acquainted.” So, to the Tweet above, a chatbot is definitely allowed to answer generic questions related to these professions as much as it likes. A chatbot answering “what are the symptoms of a heart attack” is no more acting as a professional than WebMD is by publishing an article on that.
Even if the chatbot is giving particularized advice, it probably isn’t practicing a profession because the courts still are looking for a narrow construction that avoids the first amendment issue; scholars have suggested that the right way of thinking about is whether the advisor has a fiduciary duy. Rosemond v. Markham is a good model for thinking about that in the chatbot context, holding that even if you’re giving personalized advice in response to someone’s specific question about themself, that does not necessarily establish such a relationship. (Rosemond was a columnist who gave psychological advice to people who wrote in to him, describing himself as a “family psychologist.” Someone in Kentucky complained that, because Rosemond was not a licensed professional in Kentucky, this amounted to unauthorized practice. Applying the professional speech doctrine pre-NIFLA, the court held that just responding to questions without establishing some further relationship doesn’t qualify as professional speech and thus Rosemond gets first amendment protection).
I do prefer saying “effective ban”, because I don’t think a chatbot provider could be sued for any advice it gives that doesn’t result in any actual harm. It can only be sued for the harm. Now, because medical (and legal and …) advice is a game of risk management, this means that its impractical to offer advice under those constraints.
My guess is if you look at the politicians who have been proposing this, they will also refer to this as “a ban on chatbots giving medical advice”. I haven’t looked this up, but I am a bit above 50% that the supporting side also thinks about this as a ban.
Chatbots shouldn’t claim to be a doctor, lawyer or any other licensed professional. My bill, S7263, stops chatbots from impersonating licensed professions while allowing those bots to still give advice. Here’s a thread on what the bill does/doesn’t do & why it’s important:
It’s illegal to practice high-risk professions without a license, and it’s a crime to pretend to have a license. If someone impersonates a doctor and gives advice that makes you sick, they would be held criminally liable. The same standard should apply to AI chatbots!
There’s many documented cases of chatbots giving fake license numbers. You should have the right to seek damages if a chatbot tells you it’s a doctor, a lawyer, a veterinarian, or any other licensed professional and gives you bad advice.
This legislation does not prohibit a user asking a chatbot questions or receiving general information/advice, as long as the chatbot is not presenting that information as a licensed professional. This bill does hold AI companies liable when their products harm NYers.
So to summarize, S7263:
✅ Creates liability when chatbots impersonate licensed professionals
✅ Holds chatbots to the same legal standard as humans
✅ Protects users from misinformation, scams, & fraud
S7263 does NOT:
❌ Ban chatbots from answering questions or giving advice about health, law, or any topic related to a licensed profession, as long as it is not presenting as a licensed professional
❌ Ban the use of AI for help
❌ Outlaw chatbots
It should be noted that replies on Twitter challenge her on this interpretation matching the bill text.
During AI Week I spoke to News10 to call for the passage of the New York AI Act (S1169) and the Chatbot Liabilities Bill (S7263). We must prioritize increasing transparency and accountability when using AI technologies.
I didn’t see that portion of the bill you posted, and my earlier takeaway from 5.4′s objection was that it would have permitted the kinds of chat you mention. So I think your take is largely accurate and I’m sorry for posting something like this without reading the bill all the way through; I probably wouldn’t have done so if I had understood this before making it.
If you read closely, the AI’s correction is actually only really objecting to the “related to”, which (in its interpretation) I guess is supposed to imply the idea that the bill was supposed to ban any advice “related to” medicine, engineering, etc. As Max H pointed out, there’s a reasonable read of the post that goes “bans AI from answering (some) questions related to several licensed professions like medicine...”, so the AI’s correction is assuming a reading of the above tweet that isn’t even correct.
I think the AI’s correction is designed to give the reader the impression that the bill doesn’t ban advice, without actually claiming that it doesn’t ban advice. I suspect that AI models are just not good/trustworthy enough to do this sort of thing yet. I still do think though that the tweet would be better saying “effectively ban” and removing the “related to”.
to be completely honest, i wish that your original claims had been correct
because i suspect that it wouldn’t actually make a difference at all, and that would have been a very useful way to get at the crux: that, even without this clear and direct language, an attempt to outlaw chatbots “impersonating medical professionals” would have effectively prevented AI from offering individualized medical advice even if the bill in question were not so extreme
this is just an intuition from seeing how licensure works in other areas, but it seems to be an intuition that a lot of other people in this sphere have. it would have been very useful for epistemics, imo, if we could have gotten a version of the bill that actually looked the way you originally portrayed it, and then observed whether it did in reality lead to AI no longer giving medical, legal etc advice without extensive jailbreaking
The bill itself is really short, about 500 words, a third of which are defining what a “chatbot” is[1]. I think the quality of argument would be better if people took two minutes to read the actual bill, and then a few minutes to read through sections 6512 and 6513, and articles 131, 133, 135, 136, etc of the NY education code, and then asked their favorite LLM which of the terms are terms of art.
That said, looking at the text of the bill and the referenced sections of the NY education code, I’m inclined to agree with Zvi, Eliezer, and Dean Ball. As far as I can tell, if a New Yorker describes the symptoms their dog has, and asks what home remedies are available, and the chatbot answers, and the dog dies, the New Yorker can sue for damages + legal expenses[2].
Disclaimer[3]: I am not a lawyer, and this is not legal advice.
Their definition of “chatbot” is interesting—as far as I can tell, an IVR phone tree which takes voice input (To help us route your call, please say what you are calling about. For example, if you are calling about the status of your prescription, say ‘Prescription Status’.”) counts as a “chatbot”.
A disclaimer which the bill explicitly makes unavailable to the chatbot proprietor if I’m reading correctly: “(B) A PROPRIETOR MAY NOT WAIVE OR DISCLAIM THIS LIABILITY MERELY BY NOTIFYING CONSUMERS THAT THEY ARE INTERACTING WITH A NON-HUMAN CHATBOT SYSTEM.”
It seems likely to me that the actual consequence of the bill, if enacted, map more closely to the broad version than the narrow version for ease of implementation / cost of violation reasons, and that the various tweeters share this assumption.
Maybe that’s just a differently instantiated version of the bias you’re describing, but it feels to me like a relevant distinction: They’re not literally misreading the law. They’re trying to model the consequences of its implementation.
They are literally misreading the law if the tweet says the bill “bans AI from answering questions related to several licensed professions”, and it literally doesn’t do that.
But do you see how shifting from ‘just reading’ to ‘interpreting implementation consequences’ means that the tweet may be claiming an effective ban and not a ban by the letter of the law itself? I agree that this is sloppy but it’s not out of the norm for discussion of laws in the public eye (where the consequences are foregrounded rather than the details; e.g. the stuff from like 15 years ago about hallway width in abortion clinics, which was widely reported on as ‘banning abortion clinics’, because changing hallway width was prohibitively expensive).
[Genuinely not trying to convince; it looks like you may not have seen my point yet, rather than that you’ve seen it and disagree with it, but maybe I’m wrong about that!]
But do you see how shifting from ‘just reading’ to ‘interpreting implementation consequences’ means that the tweet may be claiming an effective ban and not a ban by the letter of the law itself?
I understand what you’re saying, but that amount of charity is inappropriate. If the OP wanted to say “effective ban”, they would have done that, and then the tweet wouldn’t have misled people. And in other contexts I am almost positive that rationalists would be able to immediately register this kind of conflation as antisocial; for instance, people made similar claims that SB 1047 would “ban open source”, and several of the people mentioned above thought that was just as mendacious.
That is materially narrower than a blanket ban on AI “answering questions related to” medicine, law, dentistry, nursing, psychology, social work, engineering, etc.
So the post overstates the bill’s scope: the bill does not ban AI from answering all questions related to those professions;
Nitpick: the tweet in question does not actually use the phrases “all questions” or “blanket ban”. ChatGPT is interpreting ambiguity in a hyperbolic way to make the correction look stronger.
It’s also not clear from the excerpt of the bill that you quoted that a chatbot simply prefacing its response with “I am not a lawyer / doctor / hair stylist / etc. but”, (which would be annoying but not catastrophic) is sufficient to avoid liability.
The bill’s author is a democratic socialist that has said, among other things, “People deserve real care from real people.”, implying that the bill is indeed designed to limit what AIs can say, not just that they need to include more disclaimers when discussing certain subjects. Which is indeed very bad and stupid, even if it is not a “blanket ban” on “answering all questions” (which again, are ChatGPT’s words, not the words of anyone credible who supports or opposes the bill).
Nitpick: the tweet in question does not actually use the phrases “all questions” or “blanket ban”. ChatGPT is interpreting ambiguity in a hyperbolic way to make the correction look stronger.
I did not notice this earlier, but given Habryka’s comment, I don’t think this is a nitpick at all. I think the AI is actively trying to make the user believe that the bill doesn’t ban advice, without saying that.
Seems reasonable to me to interpret the words “ban X” as a ban on X, not as ban on some subset of X. That is certainly how people who are responding to the tweet appear to be reading it.
It’s also not clear from the excerpt of the bill that you quoted that a chatbot simply prefacing its response with “I am not a lawyer / doctor / hair stylist / etc. but”, (which would be annoying but not catastrophic) is sufficient to avoid liability.
Here are the attached sources that the AI gave, & its quotes & paraphrases from the sections:
NY State Senate Bill 2025-S7263 Summary: “Imposes liability for damages caused by a chatbot impersonating certain licensed professionals.” Sponsor memo: the bill would prohibit a chatbot from giving responses or advice that, if taken by a natural person, would constitute unauthorized practice or unauthorized use of a professional title.
STATE OF NEW YORK — S7263 bill text Section 390-f(2)(a): “A proprietor of a chatbot shall not permit such chatbot to provide any substantive response, information, or advice, or take any action which, if taken by a natural person” would constitute specified crimes of unauthorized professional practice or unauthorized use of title.
AI Chatbot Ban for Minors Passes Internet & Technology Committee, among 11 Bills The Senate press release says S7263 would “prevent AI from impersonating certain licensed professionals” and “would prohibit chatbots from giving substantive responses, including information or advice, that can be mistaken for professional counseling.”
If there is actually any ambiguity here, I am willing to bet literally anyone on this website that, if the bill goes forward, the ambiguities will be resolved in favor of a less aggressive interpretation in subsequent edits.
Seems reasonable to me to interpret the words “ban X” as a ban on X, not as ban on some subset of X. To ban something means to ban it.
The phrases “A ban on X related to Y” and “A blanket ban on all X related to Y” do not have identical meanings and interpretations in practice.
The former is ambiguous and therefore potentially misleading in the context of this bill, the latter is precise but definitely false.
I agree that the original tweet is bad and hyperbolic, but my point is that if you’re going to make a correction, you should correct the exact precise thing the person you’re correcting actually said.
If there is actually any ambiguity here, I am willing to bet literally anyone on this website that, if the bill goes forward, the ambiguities will be resolved in favor of a less aggressive interpretation in subsequent edits.
Yes, I think it is ambiguous whether this:
“would prohibit chatbots from giving substantive responses, including information or advice, that can be mistaken for professional counseling.”
means that making the chatbot include a disclaimer that it is not a licensed professional is sufficient on its own. If you start a response with “I am not a doctor, but...” and then give 2000+ words of actionable medical advice, that could reasonably be mistaken for “professional counseling”, which means it would be prohibited.
I am willing to bet literally anyone on this website that, if the bill goes forward, the ambiguities will be resolved in favor of a less aggressive interpretation in subsequent edits.
… because people look at the bill, find the places that it’s unreasonable, and contact legislators about it. The discussions about the ways the current bill is bad are how the bill gets better.
Note: Nevermind, I was right the first time.
Note: This shortform correcting a tweet is itself misleading;see this conversation between Habryka and myself.As I have pointed out before, rationalists & technology workers have a self-serving bias that makes them preferential to stories about public servants being stupid. This tweet about the NY AI bill is a great example of that:
The above post was quoted uncritically by Zvi, Eliezer, and Dean W. Ball, who each wrote some variation of “woe that state legislatures are so horrible”. The real bill is quite abit narrower than the tweet suggests; from GPT-5.4:
Public policymakers are ultimately the people who are going to be drafting laws & organizing responses about AI safety, and the reception of those laws is going to be driven partly by priors about their efficacy, developed from stuff like this. Boosting hyperbolic misinfo about particular bills, if it’s an attempt to build solidarity/shared intuitions with people in tech circles (which, to be clear, I don’t think this is), seems extremely counterproductive.
Sorry, I don’t understand, this seems like a straightforward ban:
My understanding is that giving “substantive medical advice” is “unauthorized practice” and as such is banned. That’s what the tweet above says, pretty straightforwardly. The tweet is pretty clear that it is only including “substantive responses”, so I don’t think anyone is being misled.
I would be in favor of it saying “effective ban”, but honestly, it’s so close to just a straightforward direct ban that I think the current thing is totally fine. To be clear, the bill is directly banning AI systems giving substantive medical advice. The bill is quite clear that a disclaimer would not be sufficient, and AI systems would only be allowed to discuss general health considerations. Any kind of personalized medical advice would be banned, no matter what disclaimers you use.
On disclaimers, the bill literally says:
My guess is if you look at the politicians who have been proposing this, they will also refer to this as “a ban on chatbots giving medical advice”. I haven’t looked this up, but I am a bit above 50% that the supporting side also thinks about this as a ban.
It depends on what exactly you mean by “substantive medical advice” but you have to read these statutes against the backdrop of robust free speech protections when applied to giving advice.
This is a somewhat unsettled legal area and the Supreme Court is currently considering a case (Chiles v. Salazar) that may reshape the doctrine, but if you’re talking about professional practice that purely involves speech then it gets a lot of protection. That quickly recedes once any kind of conduct is involved (i.e., any kind of physical interaction), so the government usually wins unauthorized practice cases by pointing to something other than talking that was done.
Presumptively, if you’re only talking the government can’t regulate the content of that speech without clearing an exceptionally high legal bar and so someone giving you purely verbal medical advice without any physical component is likely in the clear. That’s why it’s not unauthorized practice of medicine if you tell me you have a headache and I suggest you take an aspirin, and all of the various websites out there that let you submit a list of symptoms and then suggest a diagnosis are fine too.
To crack down on something that doesn’t involve fraud/impersonation, the government would have to rely on the so-called “professional speech” exception to the first amendment, used in some circuit courts. It’s not clear this is even good law; the Supreme Court’s majority has never endorsed it, and they cast significant doubt on in in NIFLA v. Becerra (2018).
To the extent the professional speech exception exists at all, it applies when there is a “personal nexus” between the professional and the client. The threshold inquiry is whether or not “a speaker … purport[s] to be exercising judgment on behalf of any particular individual with whose circumstances he is directly acquainted.” So, to the Tweet above, a chatbot is definitely allowed to answer generic questions related to these professions as much as it likes. A chatbot answering “what are the symptoms of a heart attack” is no more acting as a professional than WebMD is by publishing an article on that.
Even if the chatbot is giving particularized advice, it probably isn’t practicing a profession because the courts still are looking for a narrow construction that avoids the first amendment issue; scholars have suggested that the right way of thinking about is whether the advisor has a fiduciary duy. Rosemond v. Markham is a good model for thinking about that in the chatbot context, holding that even if you’re giving personalized advice in response to someone’s specific question about themself, that does not necessarily establish such a relationship. (Rosemond was a columnist who gave psychological advice to people who wrote in to him, describing himself as a “family psychologist.” Someone in Kentucky complained that, because Rosemond was not a licensed professional in Kentucky, this amounted to unauthorized practice. Applying the professional speech doctrine pre-NIFLA, the court held that just responding to questions without establishing some further relationship doesn’t qualify as professional speech and thus Rosemond gets first amendment protection).
God damnit, this makes me want to unredact my post again.
I do prefer saying “effective ban”, because I don’t think a chatbot provider could be sued for any advice it gives that doesn’t result in any actual harm. It can only be sued for the harm. Now, because medical (and legal and …) advice is a game of risk management, this means that its impractical to offer advice under those constraints.
For
I checked on Twitter.
@SenGonzalezNY (
2026-03-06) (quoting the viral tweet and attempting to clarify):It should be noted that replies on Twitter challenge her on this interpretation matching the bill text.
From a year ago there is also: @SenGonzalezNY (
2025-03-21):The co-sponsors don’t seem to have commented on this bill on Twitter.
I didn’t see that portion of the bill you posted, and my earlier takeaway from 5.4′s objection was that it would have permitted the kinds of chat you mention. So I think your take is largely accurate and I’m sorry for posting something like this without reading the bill all the way through; I probably wouldn’t have done so if I had understood this before making it.
If you read closely, the AI’s correction is actually only really objecting to the “related to”, which (in its interpretation) I guess is supposed to imply the idea that the bill was supposed to ban any advice “related to” medicine, engineering, etc. As Max H pointed out, there’s a reasonable read of the post that goes “bans AI from answering (some) questions related to several licensed professions like medicine...”, so the AI’s correction is assuming a reading of the above tweet that isn’t even correct.
I think the AI’s correction is designed to give the reader the impression that the bill doesn’t ban advice, without actually claiming that it doesn’t ban advice. I suspect that AI models are just not good/trustworthy enough to do this sort of thing yet. I still do think though that the tweet would be better saying “effectively ban” and removing the “related to”.
to be completely honest, i wish that your original claims had been correct
because i suspect that it wouldn’t actually make a difference at all, and that would have been a very useful way to get at the crux: that, even without this clear and direct language, an attempt to outlaw chatbots “impersonating medical professionals” would have effectively prevented AI from offering individualized medical advice even if the bill in question were not so extreme
this is just an intuition from seeing how licensure works in other areas, but it seems to be an intuition that a lot of other people in this sphere have. it would have been very useful for epistemics, imo, if we could have gotten a version of the bill that actually looked the way you originally portrayed it, and then observed whether it did in reality lead to AI no longer giving medical, legal etc advice without extensive jailbreaking
The bill itself is really short, about 500 words, a third of which are defining what a “chatbot” is [1] . I think the quality of argument would be better if people took two minutes to read the actual bill, and then a few minutes to read through sections 6512 and 6513, and articles 131, 133, 135, 136, etc of the NY education code, and then asked their favorite LLM which of the terms are terms of art.
That said, looking at the text of the bill and the referenced sections of the NY education code, I’m inclined to agree with Zvi, Eliezer, and Dean Ball. As far as I can tell, if a New Yorker describes the symptoms their dog has, and asks what home remedies are available, and the chatbot answers, and the dog dies, the New Yorker can sue for damages + legal expenses [2] .
Disclaimer [3] : I am not a lawyer, and this is not legal advice.
Their definition of “chatbot” is interesting—as far as I can tell, an IVR phone tree which takes voice input (To help us route your call, please say what you are calling about. For example, if you are calling about the status of your prescription, say ‘Prescription Status’.”) counts as a “chatbot”.
Maybe. Depending on whether this violation is “willful”.
A disclaimer which the bill explicitly makes unavailable to the chatbot proprietor if I’m reading correctly: “(B) A PROPRIETOR MAY NOT WAIVE OR DISCLAIM THIS LIABILITY MERELY BY NOTIFYING CONSUMERS THAT THEY ARE INTERACTING WITH A NON-HUMAN CHATBOT SYSTEM.”
It seems likely to me that the actual consequence of the bill, if enacted, map more closely to the broad version than the narrow version for ease of implementation / cost of violation reasons, and that the various tweeters share this assumption.
Maybe that’s just a differently instantiated version of the bias you’re describing, but it feels to me like a relevant distinction: They’re not literally misreading the law. They’re trying to model the consequences of its implementation.
They are literally misreading the law if the tweet says the bill “bans AI from answering questions related to several licensed professions”, and it literally doesn’t do that.
But do you see how shifting from ‘just reading’ to ‘interpreting implementation consequences’ means that the tweet may be claiming an effective ban and not a ban by the letter of the law itself? I agree that this is sloppy but it’s not out of the norm for discussion of laws in the public eye (where the consequences are foregrounded rather than the details; e.g. the stuff from like 15 years ago about hallway width in abortion clinics, which was widely reported on as ‘banning abortion clinics’, because changing hallway width was prohibitively expensive).
[Genuinely not trying to convince; it looks like you may not have seen my point yet, rather than that you’ve seen it and disagree with it, but maybe I’m wrong about that!]
I understand what you’re saying, but that amount of charity is inappropriate. If the OP wanted to say “effective ban”, they would have done that, and then the tweet wouldn’t have misled people. And in other contexts I am almost positive that rationalists would be able to immediately register this kind of conflation as antisocial; for instance, people made similar claims that SB 1047 would “ban open source”, and several of the people mentioned above thought that was just as mendacious.
Nitpick: the tweet in question does not actually use the phrases “all questions” or “blanket ban”. ChatGPT is interpreting ambiguity in a hyperbolic way to make the correction look stronger.
It’s also not clear from the excerpt of the bill that you quoted that a chatbot simply prefacing its response with “I am not a lawyer / doctor / hair stylist / etc. but”, (which would be annoying but not catastrophic) is sufficient to avoid liability.
The bill’s author is a democratic socialist that has said, among other things, “People deserve real care from real people.”, implying that the bill is indeed designed to limit what AIs can say, not just that they need to include more disclaimers when discussing certain subjects. Which is indeed very bad and stupid, even if it is not a “blanket ban” on “answering all questions” (which again, are ChatGPT’s words, not the words of anyone credible who supports or opposes the bill).
I did not notice this earlier, but given Habryka’s comment, I don’t think this is a nitpick at all. I think the AI is actively trying to make the user believe that the bill doesn’t ban advice, without saying that.
Seems reasonable to me to interpret the words “ban X” as a ban on X, not as ban on some subset of X. That is certainly how people who are responding to the tweet appear to be reading it.
Here are the attached sources that the AI gave, & its quotes & paraphrases from the sections:
NY State Senate Bill 2025-S7263
Summary: “Imposes liability for damages caused by a chatbot impersonating certain licensed professionals.” Sponsor memo: the bill would prohibit a chatbot from giving responses or advice that, if taken by a natural person, would constitute unauthorized practice or unauthorized use of a professional title.
STATE OF NEW YORK — S7263 bill text
Section 390-f(2)(a): “A proprietor of a chatbot shall not permit such chatbot to provide any substantive response, information, or advice, or take any action which, if taken by a natural person” would constitute specified crimes of unauthorized professional practice or unauthorized use of title.
AI Chatbot Ban for Minors Passes Internet & Technology Committee, among 11 Bills
The Senate press release says S7263 would “prevent AI from impersonating certain licensed professionals” and “would prohibit chatbots from giving substantive responses, including information or advice, that can be mistaken for professional counseling.”
If there is actually any ambiguity here, I am willing to bet literally anyone on this website that, if the bill goes forward, the ambiguities will be resolved in favor of a less aggressive interpretation in subsequent edits.
The phrases “A ban on X related to Y” and “A blanket ban on all X related to Y” do not have identical meanings and interpretations in practice.
The former is ambiguous and therefore potentially misleading in the context of this bill, the latter is precise but definitely false.
I agree that the original tweet is bad and hyperbolic, but my point is that if you’re going to make a correction, you should correct the exact precise thing the person you’re correcting actually said.
Yes, I think it is ambiguous whether this:
means that making the chatbot include a disclaimer that it is not a licensed professional is sufficient on its own. If you start a response with “I am not a doctor, but...” and then give 2000+ words of actionable medical advice, that could reasonably be mistaken for “professional counseling”, which means it would be prohibited.
… because people look at the bill, find the places that it’s unreasonable, and contact legislators about it. The discussions about the ways the current bill is bad are how the bill gets better.
Yeah you’re probably right.