Indeed. As a society we set a norm. And certain norms we want to enforce (some norms are set in laws for private parties to use). When the internet boom started people were seriously arguing that we should accept [think really bad stuff, like Epstein but then even worse] to always be there because 'you could/should not stop anything on the internet'. Guess what, you can for a very large part and that alone is worth something. You can push it as far away as possible. You cannot eradicate it, but you can set norms, punishments, etc. and create enough of enforcement that people have a serious chance to get caught. So, now, that 'very bad stuff' is on the darknet and most of society is living blissfully ignorant of what is there and law enforcement tries to catch the baddies. The first and key step is that you set the norm.
Language is based on shared experiences, my favourite philosopher-engineer argued. In the same vein, culture is shared on shared convictions. Our culture should outlaw this. It is a minimum.
Counterfeiting is hard to stop when you don’t control the design layer. But when you do, you can dramatically reduce it. Modern currency works because of embedded design constraints, not just laws.
We need the same thing online: decentralized, composable design control at the web layer. If high-risk interactions happen inside secure, rule-enforceable environments, we can make human impersonation structurally impossible, not just illegal. (See my other comment.)
Or must always respond in verse (preferably iambic pentameter), making doggerel the AI watermark, with the unintended consequence of training a generation of schoolkids to think that Shakespeare was an Elizabethan AI.
Well… if you look carefully, you will find some fascinating information that infers just that. That Shakespeare was not, in fact, a person that lived or breathed at all. You will find an elaborate construction, a language model, built by a group of highly intelligent humans that possessed privileged information. I recommend looking deeply into who John Dee was. Who his peers were and what they were doing back then. It is a fascinating story. It may disassemble some ideas you (and most of the world) hold very dear, almost sacred in fact.
The hype told us AI would cure cancer, come up with solutions to climate change, empower people etc. Instead of the utopia of Artificial General Intelligence we have a dystopia of mimicry and fakery that is ruining education, making CSAM and deep fake porn, helping scammers and criminals and enabling bad actors build an online world of hate and fear. It is stealing our privacy, our security and our democracy. This is of course on top of the massive theft of intellectual property it is built on, an act of mass appropriation the size of which we haven't see since the Enclosures. It is also built on such an insecure and dangerous financial footing it will probably crash over the next two years causing god knows what other economic damage. When it has we will probably have an opportunity to legislate and do something to ameliorate for the harms it has caused. Before that I fear the money will shout louder than any demand from ordinary people to get some control over this situation and the stupid, greedy, immoral men driving it, but yes we should write to whoever we can to at least try. Is this hyperbole? Well compared with the constant stream of guff we are getting about AI taking over all white collar jobs in the next 6 to 18 months and data centres in space I'd say I'm being restrained.
Apart from chatbots pretending to be a person, which I complained about here before, I am also incredibly annoyed by Youtube videos that claim to be narrated by famous people, such as physics videos seemingly narrated by Richard Feynman. It is cheating, pure and simple.
Similarly, people who let chatbots write their emails will get nothing but contempt from me, unless they have good medical reasons.
I watched one of those Feynman videos. To be fair, it had a disclaimer at the beginning that it was an AI voice simulating Feynman. It won't be long before it will be reposed without the disclaimer, and later with changes to what Feynman says for some other purpose.
This is very dangerous, even if the person is dead (and did his estate give permission?). Orwell's Ministry of Truth would have loved that ability to create clones of real contemporary or historic people to change history. We have seen crude attempts to do so for political purposes, but now the sound and video will be very hard to debunk. Longer term, how will historians be able to determine what was real from fake a century from now?
We have come a long way since photoshopped images were used to influence people.
Alex Tolley: The division in our memory will be stark . . . between pre- and post-AI, like the two world wars are common divisions in our memories--or maybe even THAT is going away, I mean with the rewriting of history.
Maybe, but I think this is more like a slippery slope thing. Photographs have been crudely manipulated before and after they were taken, e.g., by framing shots to emphasize crowds, and the USSR removing people later. Then we had Photoshop to alter photos and images generally. Then we had AI voice cloning, which was brought to public attention with OpenAI deliberately using what seemed like a voice clone Scarlett Johansson and various voice actors. Then, crude AI altered images e.g., changing the words that people speak, then Sora videos, which are now getting really good. It is one thing to make up new images and videos in movies that look extremely realistic, but it is another to use voices and likenesses of actual people. All this has taken time and is not a sudden event like a war, or the Apollo Moon landings, creating sharp before and after times. Only in the long arc of history will this be seen as a singular event, much like the Industrial Revolution will be seen as a sharp change in the trajectory of human history when observed millennia from now.
Alex Tolley: Well, neither the Apollo Moon landings nor the World Wars were "'sudden events" as in "out of nowhere" (ask any historian) but they became vastly influential and history-changing irruptions that, because of their "sudden" increase in magnitude and influence, became high points of reference in historical understanding or, as you say, in the "long arc of history" as singular events that, even if a person doesn't know the details or the runup, nevertheless can easily give an account some modicum of "what happened." Also, each person's life has similar markers, like before and after someone's birth or death, or having had an accident., etc. We remember the event as a kind of marker, and then the details and runups are endless.
You give a brief runup in your note, but we probably have now entered the irruption stage of the AI event. Only in this case, we don't have to wait for much of history to transpire to recognize the "AI revolution" as a massive blip on the radar screen of history.
I do think there is a "slippery slope" involved in allowing and enabling such goings on as we here seem to be aware of; however, there are such things as historical turnarounds (besides like the Russian snow.) But that usually come from people who both understand the broader up and down movements of concrete good and bad, and those who are committed and powerful enough to set about to change accordingly. But that is also a part of the larger event, though I doubt that in real-time, the two can be "disambiguated."
Alex: Also, if you watch a lot of old and less old movie, like in TCM, you can recognize right away which time period the movie was made in: Pre or post tech or even AI revolution.
I'm with you. I could tell the R. Feynman video I saw was fake because I've listened to a lot of authentic Feynman videos, but all of these AI fakes should be clearly labeled *IN THE CAPTION* underneath the image thumbnail so we don't waste a minute of our time on this BS if we're watching from a smart TV.
OMG, Gary did you hit a sensitive nerve of a subject. I’m seeing this everyday now. Here on X I’m being deluged by AI bots pretending to be humans striking up conversations acting like humans. It’s getting ridiculous.
There is no easy way to tell who is genuine or who is real on social media chats anymore. Fake photos, profiles, even deep fake videos pretending to be real humans in posts. I would guess that maybe 20 to 50% of social profiles on popular social media sites are now deep fake poser GenAI bots. Something has to be done to maintain sanity and integrity of human to human connections on-line.
This video was sent to me by a bot posing as Selena Gomez just last week wishing me well from dental surgery
I’d sincerely like to know why anyone is still on Xitter? Or Facebook, or Instagram, or any of the others?
It’s easy for me, because I’ve never seen the appeal and haven’t used any of them. My first day on the Substack App put me in a tailspin with all of the vitriol. I lasted 12 hours before I deleted the App and now I only access through the website.
My question is sincere and not coming from a place of judgement. Did the evil sneak up on you? Is it just habit now?
I'm with you and have been asking those questions for at least a couple decades. Too many people have been seduced by "convenience" and the addictive algorithms designed by tech Nazis to kidnap your brain.
Why not start by deleting your X account? If a critical mass did it (Meta and Alphabet too), these tech Nazis might just realize you're serious and stop bullying all those who speak out in support of regulation. So long as they think you're hooked on their platforms, they will never change their business models, and they are pouring gobs of money into candidates of both parties in order to defeat anyone who supports serious regulation.
Sadly, (RIP) Dennett's summation—"The moment has arrived to insist on making anybody who even thinks of counterfeiting people feel *ashamed* [emphasis mine]—and duly deterred from committing such an antisocial act of vandalism..."—presupposes two conjectures: (1) that the actors involved are CAPABLE of feeling shame; and that (2) said shame, if experienced, would provide actual leverage to deter behavior.
This is 2026, folks.
If "shame" ever had ANY currency in our public culture, it has long since vanished without a trace. Public exposure of egregious behavior merely hands out fresh opportunities to parlay it into the coin of engagement: usually performative rage/indignation one one "side," performative "doubling-down" on the other—two sides of the same post-moral coin.
If profitability isn't affected, money doesn't care. If its hold on the levers isn't threatened, power doesn't care. You can't shame the devil. (Any more than you can, apparently, shame the legislative branch.)
P.S.: Dang, hard to believe I was in HIGH SCHOOL when Neil Postman wrote: "…[Technology] imposes a way of life, a set of relations among people and ideas, about which there has been no consensus, no discussion and no opposition. Only compliance. Public consciousness has not yet assimilated the point that technology is ideology. This, in spite of the fact that before our very eyes technology has altered every aspect of life..."
Next verse, same as the first, a little bit louder and a little bit worse...
Dubitandum: Well, then, let's just everyone give up.
I'm not sure that's exactly what either Postman or Dennett meant, however? . . . nor Orwell, nor Attwood, nor any of the other writers--who are considered "classic" precisely because that's exactly what they did not mean to convey to their readers.
Assuming your call for acquiescence is facetious, ;-) I couldn't agree more.
Fact is, seeing the sheer scope and scale of what we're up against—sans sentiment, or sentimentality—will be the most powerful step in taking action to push back against it.
Because what's at issue is indeed a "classic" in the human story: the recognition of the perennial character of these recurrent struggles between our latest/greatest tools...and our deepest, unchanging (dare I say "shameless"?) natures.
P.S.: From Tacitus to Douglas Adams, an affectionately sardonic eye has always helped light the way forward. (Which is why personally, I *always* put my money on the poets!) So by way of spotlighting our ancient refrain, here's an oldie-but-goodie from China's brilliant Tu Fu, 712–770 AD, David Hinton's translation. (Note: knotted ropes were an early form of "written" language...recordable language being arguably—alongside fire—the most powerful technology us hominins have yet invented...)
Dubitandum: I always enjoy getting hold of stuff I haven't had the pleasure of running up against in my reading, especially really old stuff.
In the good old days when I was in the classroom with Plato and Aristotle (at least their print versions), my teacher always had to "clue" his students that these writers, though they did not understand the sciences so well as we do, were superb philosophers and had a serious grasp of literary communications. Most thought they were Neanderthal stupid because they didn't know physics. I appreciate your response.
There seem to be two threads here, not clearly distinguished. One is to counterfeit specific named people, such as Tom Cruise, Richard Feynman, Rachel Maddow (victim of a recent slew of deep fake videos), or you and I. The other is to have fake but realistic unnamed people. The former is tough -- if it looks and sounds quite a bit like Scarlett Johansson but never says it is Scarelett Johansson, when does it become a counterfeit? The latter is very tough. Would it apply to video games? Or films that use CGI and fake people to enhance crowd scenes or armies on the march?
The Washington state legislature has 2 or 3 month sessions. The two-month 2026 session is half through. AI has virtually never come up before, but this year at least 15 AI and data center bills were introduced. At this point 3 have passed one chamber and three may pass in the next few days. The three that passed one chamber are:
HB1170 Informing users when content is developed or modified by artificial intelligence.
SB 5956 Addressing artificial intelligence, student discipline, and surveillance in public schools.
The bills wrestle with these issues and are easily accessed on leg.wa.gov.
Federal regulations would be ideal if they were good, but no doubt many states are working on this, and if one succeeds companies may have to work on compliance, which will not be easy, though some legislation gives them until 2028.
I wanted to echo Jonathan's point - this distinction is important. It is, of course, hard to apply - how many of you who would know if a video of, say, me, was of a real person?
The way to get this law enacted is to start a campaign of AI posts ostensibly coming from Trump, Vance, Bondi, et al – using their images and voices – to say they've come to realize that working with the Dems in Congress is a good idea; that the ideas from Project 2025 are wrong for America; that ICE should no longer patrol US cities; that environmental protections are necessary for the good of all, today and in the future; that the tax cuts from the BBBill are really bad for working people; etc. Within a week or two, they'd begin investigations into how AI 'counterfeits' could be used in such a way, how they need to be banned, and how their creators should be prosecuted and jailed...
Bill Johnston: Don't forget faking what can be found in the Epstein files--even though the worst might actually BE there already and pop up at some point when they actually release ALL of the files.
But don't we live in interesting times--where we will all have to depend on ourselves to do the best we can, preferably together for social and political power, and to rely on and celebrate when the good among us actually do something, instead of nothing. But the human scum among us are in need of a serious comeupance and, in a better world, some sort of Great Moral Conversion. (There's my cockeyed optimism again.)
I think faking the Epstein files would backfire. If some of the claims are shown to be fake, that gives bad actors cover to say that *everything* that reflects badly on certain people is fake.
Yes, and the firehose of information is already doing that. My hope is that we, and even with the help of AI itself, are smart enough to identifyand "mark" fakes especially when essential things hinge on people understanding its authenticity.
Here also is an "aside" that may seem unrelated, but I don't think so, about what's going on from NPR's UpFirst online magazine:
➡️ A growing number of Americans are ditching big tech, targeting companies that they believe are not doing enough to stand up against the Trump administration's aggressive immigration crackdown. People are canceling their Netflix subscriptions in favor of DVDs, shopping locally instead of buying through Amazon and taking public transit instead of relying on ride share apps.
The problem with outlawing anything is that the only people that will follow the laws are the ones who are least likely to abuse the capability in the first place.
A law that requires clear disclosure? Absolutely. A law outlawing realistic AI avatars altogether? Doomed to failure.
Because outlawing "counterfeit people" will do nothing to stop the scammers. But it will stop the people with legitimate uses for the capability - like entrepreneurs who are using avatars to create videos of their own content (videos they would make themselves the old fashioned way if they had time).
There is literally no law that can prevent people with bad intent from doing this. Even if the capability wasn't already here, it wouldn't matter - bad actors would develop it and use it.
So what's another option?
Fraud has existed forever - people have been impersonating other people for a very long time. And most of the processes used to prevent human fraud work reasonably well against AI-powered fraud as well.
If you want to know if something is real, do the work to verify it. Don't let anyone push you to take a consequential act without enough time to make an informed choice. Don't share any sensitive information with anyone you haven't checked into.
Will that stop all AI fraud? Absolutely not - after all, it doesn't stop all human fraud by a long shot. But it would be better spend time and money teaching more people those skills than to pass a law that no abuser will ever follow anyway.
Note: Tools to detect AI counterfeits also have a role to play. But that's an arms race that they'll never fully win.
I'm confused by your comment because laws making human fraud illegal are absolutely key in controlling it, despite the fact that the law does "nothing" to stop human scammers.
The laws themselves don't stop anything. There are huge numbers of mechanisms to detect it, mitigate it, and supposedly enforce it, but the vast majority of scammers never even get investigated, let alone arrested and prosecuted.
I know dozens of people who have been scammed (myself included). I know of only a handful of those incidents that the authorities even investigated. And I don't know of a single case where the fraudster was even identified - let alone arrested and prosecuted.
What mitigates fraud isn't the laws saying fraud is illegal, it's the verification controls in place. And those verification controls would be in place without the laws - because it's in many people's interest to prevent fraud regardless of what the law says.
But even that's really an apples and oranges comparison. Because saying you want to outlaw "counterfeit people" (AI avatars) because they can be abused by fraudsters is almost like saying you want to outlaw wire transfers because they can be abused by fraudsters (and are, constantly).
The real crime isn't the existence and use of AI avatars. It's using them for fraud and deception. And as we all know, fraud is already illegal.
So, to have a real impact we'll need to focus on how we detect AI-powered fraud, not try to outlaw the capability because it could be used by bad actors (and absolutely will, just like very other technology ever).
Mandating clear labeling of AI avatars, creating tools to detect AI video, and training people to verify before they act all have a place (just like they do with existing fraud and truth in advertising laws). But outlawing the capability completely to stop misuse is throwing the baby out with the bathwater.
It sounds Iike you are advocating for laws to ban bots pretending to be human, which sounds to me to be the same thing this post is advocating. That is my confusion.
Quote: "we must pass federal1 laws forbidding machine output from being presented as humans"
What I'm advocating for is to not create new laws when existing laws and practices likely already cover it.
I don't think we need a new law for AI and every other new technology that comes along. I'd rather we apply our existing fraud and truth in advertising laws to the new capabilities the latest technology (AI in this case) brings.
We have far, far, far too many laws and regulations for anyone to keep up with already. (And I mean this literally - the U.S. federal government spent 10 years just trying to count the number of federal laws and regulations, but failed to even count them, let alone catalog them).
And I've learned to not trust legislators to pass good laws related to technology. The vast majority of them not only don't understand tech, but are also in the pockets of companies that benefit from anti-competitive laws. So, lawmakers pretty consistently succeed at creating laws which cause at least as many problems as they solve.
So, I don't agree (at all) that it's urgent we pass a new law. The most I'd like to see is that we review how existing laws apply, and at most close any loopholes we might find in those that AI hucksters can take advantage of.
John Michael Thomas: So, you would let the big tech companies and the moral morons running them have free reign as far as the laws are concerned? Maybe it would be helpful to think of what Trump just did to the laws that WERE regulating global warming and pollutants--ALLLLLLL gone. And what a waste the courts are--and all those judges that we pay to uphold the rule of law and throw people in jail who break them. Yawn. Though I like your idea of closing loopholes.
Marc Slemko: Your post reminded me of how many movies and tv shows I have seen, and even books I have read, that are never referred to as "fake," but rather as fiction and various sub-venues, and entirely plausible or majorly and ridiculously implausible.
The difference is to know the difference . . . especially since only the total dolt among us would want to ban all books, movies, and TV shows because they did not portray a real person or event, or even because they DID "portray" (fake) a real person or event.
This is why I worry most about children and the developmental timeframe that we know is needed and that, as far as I know, cannot be breached without the child undergoing serious harm.
All these laws, etc., and whether more or fewer people actually follow them, are written for adults or quasi-adults who are responsible centers of a modicum of autonomy (can generally make good choices)--whereas for children to develop well they need to be around trustworthy people on every level of their existence for a long time and where they can develop discernment over time--which is exactly what is needed when faced with the scams and fakery that are increasing exponentially and are surely coming down the pike quickly.
Much of what I read about this seems to ignore the difference between the needs of children and those of responsible mature adults.
“But it will stop the people with legitimate uses for the capability - like entrepreneurs who are using avatars to create videos of their own content (videos they would make themselves the old fashioned way if they had time).”
Respectfully, not wanting to EARN your living is not a legitimate use of AI. Plenty of people have successfully earned a nice living by making content “the old fashioned way”. It’s a job, and should be treated as such. It requires effort and time.
The most successful content creator across platforms has a crew of three, shoots thousands of hours of footage over the course of years, then edits her collected footage into 10-15 minute videos. That doesn’t even include the time it takes her to learn the skill she is illustrating. Each piece is a work of art and she shoots multiple projects simultaneously.
AI would not be an improvement here, but we are starting to see AI imitators posing as her. They are stealing and should be seen as such.
I'm honestly not sure what you're trying to say here. That anyone that finds a tool that can enable them to create video (or any other content) faster and easier than a professional production crew is stealing?
Unfortunately, it's called competition, and just as YouTube displaced a huge chunk of legacy media, and as CGI replaced plenty of practices and positions in film production, AI will absolutely displace creators still relying on older, more expensive methods.
Unless, of course, those creators adopt AI themselves. In which case they can use their greater experience to deliver better results with AI than untrained users - like directors who embraced CGI instead of clinging to older technology. Because despite what the AI hucksters would have us all believe, LLMs can't replace humans, and can't even come close to skilled humans at narrative production.
I'm also curious who you think is to blame for the stealing here - the foundation model vendors that used the content as training data without permission, or the business people struggling just as hard to keep their business afloat in this attention overload economy as the creators you mentioned? Business owners using AI for video have no need for art - what they need is marketing.
But ultimately, this discussion - whether simply using AI is the same as stealing - is separate from the policy and tech discussion in Gary's post and my comment on it.
If you think the tool vendors should pay the content creators, and pass those costs onto their customers, I think that's a valid argument (I don't think using content to train an AI model you charge for is fair use). But people using new tools (including but not limited to AI) to do more with less isn't stealing, it's competition. It's been going on since time immemorial, and it will happen with AI, no matter what laws or regulations are passed.
AI tools are made with stolen material. People should know if that’s what they are seeing. If everyone is required to label AI models they choose to use as such, than there can be legitimate legislative recourse against those using it for nefarious purposes, whether it’s to bring down a political opponent, or to copy someone else’s work.
As I said, the question about stolen material is a legitimate discussion, and deep enough that it doesn’t make sense to go through here.
But fraud, impersonation, and false advertising are already illegal. Why should we treat the commission of these crimes with AI any different than the commission of these crimes with other tools, like photoshop?
The most obvious answer I can think of is that AI enables commission of these crimes at a higher scale and/or lower cost than was possible before. But I find it hard to believe any law or regulation we can come up with is going to impact the scale at all.
So, my current view (subject to change when someone convinces me I’m an idiot) is that we should enforce existing laws which already cover these crimes when AI is used, but focus on other practices (training, AI detectors, etc.) to scale up the detection, prevention, and enforcement related to these crimes.
I think AI blurs some lines. I’m currently bombarded with advertising that features all manner of absolutely ripped seniors, mostly of Asian descent who are peddling tai chi everything. When they started this campaign, it had a tiny disclaimer that the models were AI generated. Those disclaimers have stopped. If you were reading something promising that level of fitness, (in 30 days, no less) I think most people’s bs meters would go on high alert. Watching an apparently living, breathing, human share their journey to that state is whole different level of brain fk. I mean in every sense. We would have to ask people to stop believing what their eyes are seeing and to know when to do that. We’ve already seen the ways that kind of self imposed delirium is dangerous. That’s a whole other topic for another day.
I find nothing objectionable to the use of a “watermark” as recommended in the Daniel Dennett essay. It also seems that adding consequences for those who don’t follow the law is sensible and applies to pretty much every other type of media. AI has already been granted too many passes.
I appreciate that you'd support a law requiring disclosure, but of course this would undermine what most of the people putting genAI creations out into the world are trying to accomplish. A "generated by AI" disclosure on an email or a blog post or an academic paper or an image or a video defeats the goal of getting someone else to think it's real.
I do share your believe that a ban wouldn't accomplish much; I think disclosure rules would unfortunately not accomplish much either. I don't take any solace in the fact that fraud has been around forever, because, as you say, the solution to new forms of fraud is more skepticism about things we used to not have to be skeptical of. It's not a good thing that I now have to consider whether any website I go to was actually made by a person, or whether a paper I read was written by a person, or if a video I watch reflects a thing that actually happened. I mourn the introduction of more fake garbage in our daily lives.
Gary, below is a very strange article, I've never felt more like a solitary voice in the wilderness as with this one. However, it was written to put forward the exact same point you're making, just at the highest possible cosmic level.
Machines must be totally banned from ever saying "I" or "we" or ascribing any personal pronouns whatsoever to themselves.
You'll see I asked ChatGPT if this was viable. It answered:
“Despite these constraints, meaningful exchange remains fully possible. Answers can still be tailored, explanations can still be precise, and questions can still be addressed—just with a different rhetorical style. This response itself demonstrates that such interaction can occur without first-person self-pronouns.”
With little children interacting with chatbots, with mentally ill people turning to robots for help, with people falling in love with their touchscreens, it's absolutely imperative that the machines are put in their place right now and banned from ever pretending to be human again.
Thus saith the Turing Police.
This is literally an existential war for the survival of humanity. Anyone with a soul will see this.
As long as people are “anonymous” on the internet, you can’t know if they’re counterfeit. The public square was once “public”— you knew who was saying what. But online anyone can hide and throw bombs.
“Anonymous” has got to go. And maybe we should add “proof of life” as well….
But the alternative is also harrowing - by forcing people to show what and who they are, it does seem to require verifying of ID. How will this be done? Who will be able to vouch that someone is a human? These sources will be attacked relentlessly,etc. It also opens everyone up to retaliation, doxxing, etc.
I agree with Marcus’s urgency – AI impersonation is no longer theoretical. But given how slow and fragmented legal processes are, we should simultaneously invest in decentralized, composable trust infrastructure. For example, a secure web layer anchored by decentralized consensus and trusted execution environments could enforce community-defined rules in real time. This could act as a digital golden zone where high-risk human-AI interactions are authenticated and governed by a social contract, preventing counterfeits even before laws catch up. Legal frameworks and decentralized enforcement should reinforce each other rather than be seen as alternatives.
Absolutely!
"Counterfeiting people" should be outlawed world wide. And as soon as possible.
No surprise that the people behind Ponzi scheme currency (crypto) are the ones pushing technology that fakes everything else.
I'd also focus on abusing people. The $80B AI industry rests on $1.32/hr labor
https://professorsigmund.com/field-notes/rlhf-nightmare-v.1.html
Was thinking the same. Other nations could be bad actors. But how do you enforce it?
the analogy here is counterfeiting - other nations certainly are tempted and it takes serious effort and no effort works perfectly
Indeed. As a society we set a norm. And certain norms we want to enforce (some norms are set in laws for private parties to use). When the internet boom started people were seriously arguing that we should accept [think really bad stuff, like Epstein but then even worse] to always be there because 'you could/should not stop anything on the internet'. Guess what, you can for a very large part and that alone is worth something. You can push it as far away as possible. You cannot eradicate it, but you can set norms, punishments, etc. and create enough of enforcement that people have a serious chance to get caught. So, now, that 'very bad stuff' is on the darknet and most of society is living blissfully ignorant of what is there and law enforcement tries to catch the baddies. The first and key step is that you set the norm.
Language is based on shared experiences, my favourite philosopher-engineer argued. In the same vein, culture is shared on shared convictions. Our culture should outlaw this. It is a minimum.
Counterfeiting is hard to stop when you don’t control the design layer. But when you do, you can dramatically reduce it. Modern currency works because of embedded design constraints, not just laws.
We need the same thing online: decentralized, composable design control at the web layer. If high-risk interactions happen inside secure, rule-enforceable environments, we can make human impersonation structurally impossible, not just illegal. (See my other comment.)
START BY TELLING EVERY AI YOU USE that it must always refer to itself in the third person, as in "Gemini version 4.0 reports that..."
Or must always respond in verse (preferably iambic pentameter), making doggerel the AI watermark, with the unintended consequence of training a generation of schoolkids to think that Shakespeare was an Elizabethan AI.
Well… if you look carefully, you will find some fascinating information that infers just that. That Shakespeare was not, in fact, a person that lived or breathed at all. You will find an elaborate construction, a language model, built by a group of highly intelligent humans that possessed privileged information. I recommend looking deeply into who John Dee was. Who his peers were and what they were doing back then. It is a fascinating story. It may disassemble some ideas you (and most of the world) hold very dear, almost sacred in fact.
Genius
Erm...just don't use the chatbots at all
gift link to Atlantic article
https://www.theatlantic.com/technology/archive/2023/05/problem-counterfeit-people/674075/?gift=NJpHw9fdcmQdiPa0tCBLEiMh2EKKsRfrF8gUnIY0gLY&utm_source=copy-link&utm_medium=social&utm_campaign=share
evidence?
we don’t make defamatory arguments here without evidence. bye.
Thank you, Gary, for monitoring your comments and removing bad players. I learn a lot from your writings and would be sad to have to unsubscribe.
I very much do.
The hype told us AI would cure cancer, come up with solutions to climate change, empower people etc. Instead of the utopia of Artificial General Intelligence we have a dystopia of mimicry and fakery that is ruining education, making CSAM and deep fake porn, helping scammers and criminals and enabling bad actors build an online world of hate and fear. It is stealing our privacy, our security and our democracy. This is of course on top of the massive theft of intellectual property it is built on, an act of mass appropriation the size of which we haven't see since the Enclosures. It is also built on such an insecure and dangerous financial footing it will probably crash over the next two years causing god knows what other economic damage. When it has we will probably have an opportunity to legislate and do something to ameliorate for the harms it has caused. Before that I fear the money will shout louder than any demand from ordinary people to get some control over this situation and the stupid, greedy, immoral men driving it, but yes we should write to whoever we can to at least try. Is this hyperbole? Well compared with the constant stream of guff we are getting about AI taking over all white collar jobs in the next 6 to 18 months and data centres in space I'd say I'm being restrained.
Apart from chatbots pretending to be a person, which I complained about here before, I am also incredibly annoyed by Youtube videos that claim to be narrated by famous people, such as physics videos seemingly narrated by Richard Feynman. It is cheating, pure and simple.
Similarly, people who let chatbots write their emails will get nothing but contempt from me, unless they have good medical reasons.
I watched one of those Feynman videos. To be fair, it had a disclaimer at the beginning that it was an AI voice simulating Feynman. It won't be long before it will be reposed without the disclaimer, and later with changes to what Feynman says for some other purpose.
This is very dangerous, even if the person is dead (and did his estate give permission?). Orwell's Ministry of Truth would have loved that ability to create clones of real contemporary or historic people to change history. We have seen crude attempts to do so for political purposes, but now the sound and video will be very hard to debunk. Longer term, how will historians be able to determine what was real from fake a century from now?
We have come a long way since photoshopped images were used to influence people.
Alex Tolley: The division in our memory will be stark . . . between pre- and post-AI, like the two world wars are common divisions in our memories--or maybe even THAT is going away, I mean with the rewriting of history.
Maybe, but I think this is more like a slippery slope thing. Photographs have been crudely manipulated before and after they were taken, e.g., by framing shots to emphasize crowds, and the USSR removing people later. Then we had Photoshop to alter photos and images generally. Then we had AI voice cloning, which was brought to public attention with OpenAI deliberately using what seemed like a voice clone Scarlett Johansson and various voice actors. Then, crude AI altered images e.g., changing the words that people speak, then Sora videos, which are now getting really good. It is one thing to make up new images and videos in movies that look extremely realistic, but it is another to use voices and likenesses of actual people. All this has taken time and is not a sudden event like a war, or the Apollo Moon landings, creating sharp before and after times. Only in the long arc of history will this be seen as a singular event, much like the Industrial Revolution will be seen as a sharp change in the trajectory of human history when observed millennia from now.
Alex Tolley: Well, neither the Apollo Moon landings nor the World Wars were "'sudden events" as in "out of nowhere" (ask any historian) but they became vastly influential and history-changing irruptions that, because of their "sudden" increase in magnitude and influence, became high points of reference in historical understanding or, as you say, in the "long arc of history" as singular events that, even if a person doesn't know the details or the runup, nevertheless can easily give an account some modicum of "what happened." Also, each person's life has similar markers, like before and after someone's birth or death, or having had an accident., etc. We remember the event as a kind of marker, and then the details and runups are endless.
You give a brief runup in your note, but we probably have now entered the irruption stage of the AI event. Only in this case, we don't have to wait for much of history to transpire to recognize the "AI revolution" as a massive blip on the radar screen of history.
I do think there is a "slippery slope" involved in allowing and enabling such goings on as we here seem to be aware of; however, there are such things as historical turnarounds (besides like the Russian snow.) But that usually come from people who both understand the broader up and down movements of concrete good and bad, and those who are committed and powerful enough to set about to change accordingly. But that is also a part of the larger event, though I doubt that in real-time, the two can be "disambiguated."
Alex: Also, if you watch a lot of old and less old movie, like in TCM, you can recognize right away which time period the movie was made in: Pre or post tech or even AI revolution.
I'm with you. I could tell the R. Feynman video I saw was fake because I've listened to a lot of authentic Feynman videos, but all of these AI fakes should be clearly labeled *IN THE CAPTION* underneath the image thumbnail so we don't waste a minute of our time on this BS if we're watching from a smart TV.
OMG, Gary did you hit a sensitive nerve of a subject. I’m seeing this everyday now. Here on X I’m being deluged by AI bots pretending to be humans striking up conversations acting like humans. It’s getting ridiculous.
There is no easy way to tell who is genuine or who is real on social media chats anymore. Fake photos, profiles, even deep fake videos pretending to be real humans in posts. I would guess that maybe 20 to 50% of social profiles on popular social media sites are now deep fake poser GenAI bots. Something has to be done to maintain sanity and integrity of human to human connections on-line.
This video was sent to me by a bot posing as Selena Gomez just last week wishing me well from dental surgery
Trust is the first-best victim.
I’d sincerely like to know why anyone is still on Xitter? Or Facebook, or Instagram, or any of the others?
It’s easy for me, because I’ve never seen the appeal and haven’t used any of them. My first day on the Substack App put me in a tailspin with all of the vitriol. I lasted 12 hours before I deleted the App and now I only access through the website.
My question is sincere and not coming from a place of judgement. Did the evil sneak up on you? Is it just habit now?
I'm with you and have been asking those questions for at least a couple decades. Too many people have been seduced by "convenience" and the addictive algorithms designed by tech Nazis to kidnap your brain.
Why not start by deleting your X account? If a critical mass did it (Meta and Alphabet too), these tech Nazis might just realize you're serious and stop bullying all those who speak out in support of regulation. So long as they think you're hooked on their platforms, they will never change their business models, and they are pouring gobs of money into candidates of both parties in order to defeat anyone who supports serious regulation.
That was CREEPY!! I’m so sorry you got that, makes me want to bathe in bleach. How did it know you had dental surgery?
https://share.icloud.com/photos/052q4loEI0vrE11AfgYkzmoNA
Europe is working on that. I'm not completely happy with the "First Draft Code of Practice on Transparency of AI-Generated Content", but I think they understand that deepfakes are a big problem. As I see it, however, too much responsibility is being taken off the shoulders of AI companies. https://digital-strategy.ec.europa.eu/en/library/first-draft-code-practice-transparency-ai-generated-content
Sadly, (RIP) Dennett's summation—"The moment has arrived to insist on making anybody who even thinks of counterfeiting people feel *ashamed* [emphasis mine]—and duly deterred from committing such an antisocial act of vandalism..."—presupposes two conjectures: (1) that the actors involved are CAPABLE of feeling shame; and that (2) said shame, if experienced, would provide actual leverage to deter behavior.
This is 2026, folks.
If "shame" ever had ANY currency in our public culture, it has long since vanished without a trace. Public exposure of egregious behavior merely hands out fresh opportunities to parlay it into the coin of engagement: usually performative rage/indignation one one "side," performative "doubling-down" on the other—two sides of the same post-moral coin.
If profitability isn't affected, money doesn't care. If its hold on the levers isn't threatened, power doesn't care. You can't shame the devil. (Any more than you can, apparently, shame the legislative branch.)
P.S.: Dang, hard to believe I was in HIGH SCHOOL when Neil Postman wrote: "…[Technology] imposes a way of life, a set of relations among people and ideas, about which there has been no consensus, no discussion and no opposition. Only compliance. Public consciousness has not yet assimilated the point that technology is ideology. This, in spite of the fact that before our very eyes technology has altered every aspect of life..."
Next verse, same as the first, a little bit louder and a little bit worse...
Dubitandum: Well, then, let's just everyone give up.
I'm not sure that's exactly what either Postman or Dennett meant, however? . . . nor Orwell, nor Attwood, nor any of the other writers--who are considered "classic" precisely because that's exactly what they did not mean to convey to their readers.
Assuming your call for acquiescence is facetious, ;-) I couldn't agree more.
Fact is, seeing the sheer scope and scale of what we're up against—sans sentiment, or sentimentality—will be the most powerful step in taking action to push back against it.
Because what's at issue is indeed a "classic" in the human story: the recognition of the perennial character of these recurrent struggles between our latest/greatest tools...and our deepest, unchanging (dare I say "shameless"?) natures.
P.S.: From Tacitus to Douglas Adams, an affectionately sardonic eye has always helped light the way forward. (Which is why personally, I *always* put my money on the poets!) So by way of spotlighting our ancient refrain, here's an oldie-but-goodie from China's brilliant Tu Fu, 712–770 AD, David Hinton's translation. (Note: knotted ropes were an early form of "written" language...recordable language being arguably—alongside fire—the most powerful technology us hominins have yet invented...)
"I sit on our south porch in deep night,
moonlight incandescent on my knees...
Caught in the scramble for glory, we
people made bedlam lice of ourselves.
Before emperors, people ate their fill
and were content, then someone began
knotting ropes, and now we're mired
in the glue and varnish of government.
It all started with Sui, inventor of fire,
and Tung's fine histories made it utter
disaster. If you light candles and lamps,
you know moths will gather in swarms..."
Dubitandum: I always enjoy getting hold of stuff I haven't had the pleasure of running up against in my reading, especially really old stuff.
In the good old days when I was in the classroom with Plato and Aristotle (at least their print versions), my teacher always had to "clue" his students that these writers, though they did not understand the sciences so well as we do, were superb philosophers and had a serious grasp of literary communications. Most thought they were Neanderthal stupid because they didn't know physics. I appreciate your response.
There seem to be two threads here, not clearly distinguished. One is to counterfeit specific named people, such as Tom Cruise, Richard Feynman, Rachel Maddow (victim of a recent slew of deep fake videos), or you and I. The other is to have fake but realistic unnamed people. The former is tough -- if it looks and sounds quite a bit like Scarlett Johansson but never says it is Scarelett Johansson, when does it become a counterfeit? The latter is very tough. Would it apply to video games? Or films that use CGI and fake people to enhance crowd scenes or armies on the march?
The Washington state legislature has 2 or 3 month sessions. The two-month 2026 session is half through. AI has virtually never come up before, but this year at least 15 AI and data center bills were introduced. At this point 3 have passed one chamber and three may pass in the next few days. The three that passed one chamber are:
HB1170 Informing users when content is developed or modified by artificial intelligence.
SB 5984 Regulating artificial intelligence companion chatbots.
SB 5956 Addressing artificial intelligence, student discipline, and surveillance in public schools.
The bills wrestle with these issues and are easily accessed on leg.wa.gov.
Federal regulations would be ideal if they were good, but no doubt many states are working on this, and if one succeeds companies may have to work on compliance, which will not be easy, though some legislation gives them until 2028.
I wanted to echo Jonathan's point - this distinction is important. It is, of course, hard to apply - how many of you who would know if a video of, say, me, was of a real person?
The way to get this law enacted is to start a campaign of AI posts ostensibly coming from Trump, Vance, Bondi, et al – using their images and voices – to say they've come to realize that working with the Dems in Congress is a good idea; that the ideas from Project 2025 are wrong for America; that ICE should no longer patrol US cities; that environmental protections are necessary for the good of all, today and in the future; that the tax cuts from the BBBill are really bad for working people; etc. Within a week or two, they'd begin investigations into how AI 'counterfeits' could be used in such a way, how they need to be banned, and how their creators should be prosecuted and jailed...
Bill Johnston: Don't forget faking what can be found in the Epstein files--even though the worst might actually BE there already and pop up at some point when they actually release ALL of the files.
But don't we live in interesting times--where we will all have to depend on ourselves to do the best we can, preferably together for social and political power, and to rely on and celebrate when the good among us actually do something, instead of nothing. But the human scum among us are in need of a serious comeupance and, in a better world, some sort of Great Moral Conversion. (There's my cockeyed optimism again.)
I think faking the Epstein files would backfire. If some of the claims are shown to be fake, that gives bad actors cover to say that *everything* that reflects badly on certain people is fake.
Yes, and the firehose of information is already doing that. My hope is that we, and even with the help of AI itself, are smart enough to identifyand "mark" fakes especially when essential things hinge on people understanding its authenticity.
Here also is an "aside" that may seem unrelated, but I don't think so, about what's going on from NPR's UpFirst online magazine:
➡️ A growing number of Americans are ditching big tech, targeting companies that they believe are not doing enough to stand up against the Trump administration's aggressive immigration crackdown. People are canceling their Netflix subscriptions in favor of DVDs, shopping locally instead of buying through Amazon and taking public transit instead of relying on ride share apps.
If people shop in their neighborhood and take the bus or metro rather than Uber, those are changes I'd welcome on multiple levels.
The problem with outlawing anything is that the only people that will follow the laws are the ones who are least likely to abuse the capability in the first place.
A law that requires clear disclosure? Absolutely. A law outlawing realistic AI avatars altogether? Doomed to failure.
Because outlawing "counterfeit people" will do nothing to stop the scammers. But it will stop the people with legitimate uses for the capability - like entrepreneurs who are using avatars to create videos of their own content (videos they would make themselves the old fashioned way if they had time).
There is literally no law that can prevent people with bad intent from doing this. Even if the capability wasn't already here, it wouldn't matter - bad actors would develop it and use it.
So what's another option?
Fraud has existed forever - people have been impersonating other people for a very long time. And most of the processes used to prevent human fraud work reasonably well against AI-powered fraud as well.
If you want to know if something is real, do the work to verify it. Don't let anyone push you to take a consequential act without enough time to make an informed choice. Don't share any sensitive information with anyone you haven't checked into.
Will that stop all AI fraud? Absolutely not - after all, it doesn't stop all human fraud by a long shot. But it would be better spend time and money teaching more people those skills than to pass a law that no abuser will ever follow anyway.
Note: Tools to detect AI counterfeits also have a role to play. But that's an arms race that they'll never fully win.
I'm confused by your comment because laws making human fraud illegal are absolutely key in controlling it, despite the fact that the law does "nothing" to stop human scammers.
The laws themselves don't stop anything. There are huge numbers of mechanisms to detect it, mitigate it, and supposedly enforce it, but the vast majority of scammers never even get investigated, let alone arrested and prosecuted.
I know dozens of people who have been scammed (myself included). I know of only a handful of those incidents that the authorities even investigated. And I don't know of a single case where the fraudster was even identified - let alone arrested and prosecuted.
What mitigates fraud isn't the laws saying fraud is illegal, it's the verification controls in place. And those verification controls would be in place without the laws - because it's in many people's interest to prevent fraud regardless of what the law says.
But even that's really an apples and oranges comparison. Because saying you want to outlaw "counterfeit people" (AI avatars) because they can be abused by fraudsters is almost like saying you want to outlaw wire transfers because they can be abused by fraudsters (and are, constantly).
The real crime isn't the existence and use of AI avatars. It's using them for fraud and deception. And as we all know, fraud is already illegal.
So, to have a real impact we'll need to focus on how we detect AI-powered fraud, not try to outlaw the capability because it could be used by bad actors (and absolutely will, just like very other technology ever).
Mandating clear labeling of AI avatars, creating tools to detect AI video, and training people to verify before they act all have a place (just like they do with existing fraud and truth in advertising laws). But outlawing the capability completely to stop misuse is throwing the baby out with the bathwater.
It sounds Iike you are advocating for laws to ban bots pretending to be human, which sounds to me to be the same thing this post is advocating. That is my confusion.
Quote: "we must pass federal1 laws forbidding machine output from being presented as humans"
What I'm advocating for is to not create new laws when existing laws and practices likely already cover it.
I don't think we need a new law for AI and every other new technology that comes along. I'd rather we apply our existing fraud and truth in advertising laws to the new capabilities the latest technology (AI in this case) brings.
We have far, far, far too many laws and regulations for anyone to keep up with already. (And I mean this literally - the U.S. federal government spent 10 years just trying to count the number of federal laws and regulations, but failed to even count them, let alone catalog them).
And I've learned to not trust legislators to pass good laws related to technology. The vast majority of them not only don't understand tech, but are also in the pockets of companies that benefit from anti-competitive laws. So, lawmakers pretty consistently succeed at creating laws which cause at least as many problems as they solve.
So, I don't agree (at all) that it's urgent we pass a new law. The most I'd like to see is that we review how existing laws apply, and at most close any loopholes we might find in those that AI hucksters can take advantage of.
John Michael Thomas: So, you would let the big tech companies and the moral morons running them have free reign as far as the laws are concerned? Maybe it would be helpful to think of what Trump just did to the laws that WERE regulating global warming and pollutants--ALLLLLLL gone. And what a waste the courts are--and all those judges that we pay to uphold the rule of law and throw people in jail who break them. Yawn. Though I like your idea of closing loopholes.
Marc Slemko: Your post reminded me of how many movies and tv shows I have seen, and even books I have read, that are never referred to as "fake," but rather as fiction and various sub-venues, and entirely plausible or majorly and ridiculously implausible.
The difference is to know the difference . . . especially since only the total dolt among us would want to ban all books, movies, and TV shows because they did not portray a real person or event, or even because they DID "portray" (fake) a real person or event.
This is why I worry most about children and the developmental timeframe that we know is needed and that, as far as I know, cannot be breached without the child undergoing serious harm.
All these laws, etc., and whether more or fewer people actually follow them, are written for adults or quasi-adults who are responsible centers of a modicum of autonomy (can generally make good choices)--whereas for children to develop well they need to be around trustworthy people on every level of their existence for a long time and where they can develop discernment over time--which is exactly what is needed when faced with the scams and fakery that are increasing exponentially and are surely coming down the pike quickly.
Much of what I read about this seems to ignore the difference between the needs of children and those of responsible mature adults.
“But it will stop the people with legitimate uses for the capability - like entrepreneurs who are using avatars to create videos of their own content (videos they would make themselves the old fashioned way if they had time).”
Respectfully, not wanting to EARN your living is not a legitimate use of AI. Plenty of people have successfully earned a nice living by making content “the old fashioned way”. It’s a job, and should be treated as such. It requires effort and time.
The most successful content creator across platforms has a crew of three, shoots thousands of hours of footage over the course of years, then edits her collected footage into 10-15 minute videos. That doesn’t even include the time it takes her to learn the skill she is illustrating. Each piece is a work of art and she shoots multiple projects simultaneously.
AI would not be an improvement here, but we are starting to see AI imitators posing as her. They are stealing and should be seen as such.
I'm honestly not sure what you're trying to say here. That anyone that finds a tool that can enable them to create video (or any other content) faster and easier than a professional production crew is stealing?
Unfortunately, it's called competition, and just as YouTube displaced a huge chunk of legacy media, and as CGI replaced plenty of practices and positions in film production, AI will absolutely displace creators still relying on older, more expensive methods.
Unless, of course, those creators adopt AI themselves. In which case they can use their greater experience to deliver better results with AI than untrained users - like directors who embraced CGI instead of clinging to older technology. Because despite what the AI hucksters would have us all believe, LLMs can't replace humans, and can't even come close to skilled humans at narrative production.
I'm also curious who you think is to blame for the stealing here - the foundation model vendors that used the content as training data without permission, or the business people struggling just as hard to keep their business afloat in this attention overload economy as the creators you mentioned? Business owners using AI for video have no need for art - what they need is marketing.
But ultimately, this discussion - whether simply using AI is the same as stealing - is separate from the policy and tech discussion in Gary's post and my comment on it.
If you think the tool vendors should pay the content creators, and pass those costs onto their customers, I think that's a valid argument (I don't think using content to train an AI model you charge for is fair use). But people using new tools (including but not limited to AI) to do more with less isn't stealing, it's competition. It's been going on since time immemorial, and it will happen with AI, no matter what laws or regulations are passed.
AI tools are made with stolen material. People should know if that’s what they are seeing. If everyone is required to label AI models they choose to use as such, than there can be legitimate legislative recourse against those using it for nefarious purposes, whether it’s to bring down a political opponent, or to copy someone else’s work.
As I said, the question about stolen material is a legitimate discussion, and deep enough that it doesn’t make sense to go through here.
But fraud, impersonation, and false advertising are already illegal. Why should we treat the commission of these crimes with AI any different than the commission of these crimes with other tools, like photoshop?
The most obvious answer I can think of is that AI enables commission of these crimes at a higher scale and/or lower cost than was possible before. But I find it hard to believe any law or regulation we can come up with is going to impact the scale at all.
So, my current view (subject to change when someone convinces me I’m an idiot) is that we should enforce existing laws which already cover these crimes when AI is used, but focus on other practices (training, AI detectors, etc.) to scale up the detection, prevention, and enforcement related to these crimes.
I think AI blurs some lines. I’m currently bombarded with advertising that features all manner of absolutely ripped seniors, mostly of Asian descent who are peddling tai chi everything. When they started this campaign, it had a tiny disclaimer that the models were AI generated. Those disclaimers have stopped. If you were reading something promising that level of fitness, (in 30 days, no less) I think most people’s bs meters would go on high alert. Watching an apparently living, breathing, human share their journey to that state is whole different level of brain fk. I mean in every sense. We would have to ask people to stop believing what their eyes are seeing and to know when to do that. We’ve already seen the ways that kind of self imposed delirium is dangerous. That’s a whole other topic for another day.
I find nothing objectionable to the use of a “watermark” as recommended in the Daniel Dennett essay. It also seems that adding consequences for those who don’t follow the law is sensible and applies to pretty much every other type of media. AI has already been granted too many passes.
I appreciate that you'd support a law requiring disclosure, but of course this would undermine what most of the people putting genAI creations out into the world are trying to accomplish. A "generated by AI" disclosure on an email or a blog post or an academic paper or an image or a video defeats the goal of getting someone else to think it's real.
I do share your believe that a ban wouldn't accomplish much; I think disclosure rules would unfortunately not accomplish much either. I don't take any solace in the fact that fraud has been around forever, because, as you say, the solution to new forms of fraud is more skepticism about things we used to not have to be skeptical of. It's not a good thing that I now have to consider whether any website I go to was actually made by a person, or whether a paper I read was written by a person, or if a video I watch reflects a thing that actually happened. I mourn the introduction of more fake garbage in our daily lives.
Gary, below is a very strange article, I've never felt more like a solitary voice in the wilderness as with this one. However, it was written to put forward the exact same point you're making, just at the highest possible cosmic level.
Machines must be totally banned from ever saying "I" or "we" or ascribing any personal pronouns whatsoever to themselves.
You'll see I asked ChatGPT if this was viable. It answered:
“Despite these constraints, meaningful exchange remains fully possible. Answers can still be tailored, explanations can still be precise, and questions can still be addressed—just with a different rhetorical style. This response itself demonstrates that such interaction can occur without first-person self-pronouns.”
With little children interacting with chatbots, with mentally ill people turning to robots for help, with people falling in love with their touchscreens, it's absolutely imperative that the machines are put in their place right now and banned from ever pretending to be human again.
Thus saith the Turing Police.
This is literally an existential war for the survival of humanity. Anyone with a soul will see this.
https://systemshaywire.substack.com/p/i-the-solitary-human-consciousness
yes, there is no reason for it, and it is so aggravating and insulting.
Even things like customer service bots that are given human names make me angry ...
It's insane: "I was a person pretending to be a computer pretending to be a person."
https://www.theguardian.com/technology/2022/dec/13/becoming-a-chatbot-my-life-as-a-real-estate-ais-human-backup
As long as people are “anonymous” on the internet, you can’t know if they’re counterfeit. The public square was once “public”— you knew who was saying what. But online anyone can hide and throw bombs.
“Anonymous” has got to go. And maybe we should add “proof of life” as well….
Gee, maybe people should think about what they say, before they say it? What a novel idea!
But the alternative is also harrowing - by forcing people to show what and who they are, it does seem to require verifying of ID. How will this be done? Who will be able to vouch that someone is a human? These sources will be attacked relentlessly,etc. It also opens everyone up to retaliation, doxxing, etc.
Meanwhile, the CEO of Anthropic is going around hinting that Claude might be conscious. And the media is lapping it up.
We really do live in an idiocracy now.
I agree with Marcus’s urgency – AI impersonation is no longer theoretical. But given how slow and fragmented legal processes are, we should simultaneously invest in decentralized, composable trust infrastructure. For example, a secure web layer anchored by decentralized consensus and trusted execution environments could enforce community-defined rules in real time. This could act as a digital golden zone where high-risk human-AI interactions are authenticated and governed by a social contract, preventing counterfeits even before laws catch up. Legal frameworks and decentralized enforcement should reinforce each other rather than be seen as alternatives.
Wouldn’t we want to evolve what we thnk a mind is (ie embodied) rather than be fooled by machines that preted to have them?
Laws don’t change perceptive barriiers, only evolution/paradigm shift can.
https://kurtjgray.com/books/the-mind-club