After reading this essay, I’m left wondering: What do you think Anthropic should have done with Mythos?
Like, I agree that “tread carefully” is certainly good advice here, but it’s not clear *how* they could’ve tread more carefully. Outside of vibes-y actions like verbally appeasing the Trump admin or hiring a few republicans, I don’t k know what you would have recommended policy-wise. Sit on frontier models until after OpenAI releases a better one? That would cost them the lead in the AI race forever. Run the release by the Trump admin first? What else were Glasswing, the NSA and Pentagon briefings, etc? Add more safety measures? To shut down all GPT-5.5 levels of cyber capabilities *with near-perfect reliability* probably isn’t technically feasible.
The only thing I can think of that might have realistically worked would be waiting for the government to set up their EO cybersecurity clearinghouse, going through it, and then releasing Fable. But I’m not convinced that would’ve been enough. There’s a high chance that the jailbreak would’ve gone unnoticed (I’m not sure the clearinghouse would be Amazon-level right off the bat), and a high chance that the admin would do their inspection, approve it, and then jump on Anthropic anyway as soon as someone finds a partial jailbreak—because they wanted to.
I agree that Dario et al need to take a few levels in realpolitik, but the amount of blame directed at Anthropic feels a bit like “What was she wearing?”
well, look. Anthropic's CEO made inflammatory comments about the President on his personal social media account. He hired many of the senior architects of Biden's AI policy. His company lobbied against the moratorium, and he wrote an oped about it, and this was a priority of some in the admin. His company also lobbied for the GAIN AI Act, a bill the admin did not like, and similarly, Dario wrote an essay about the importance of export controls. Dario made statements about AI and labor that the administration perceived as alarmist and outside his wheelhouse. None of these things advanced the strategic interests of Anthropic, near as I can tell, and all of them antagonized the administration. Any one of them, maybe, is fine, but the combination of them is a pattern of behavior that causes the administration, not without good reason, to see Anthropic as an enemy. And this pattern of behavior also burned political capital that would be coming in handy around the time of the DoW dispute. Ultimately the government is abusing its power, and I think I have been clearer than almost any other Republican about that fact. But Anthropic is behaving as though its rights of free speech also confer a right to infinite political capital. The two are not the same.
Fair point about the broader history of Anthropic. It does look a bit different when you put everything together.
On the flip side, though, it feels like conceding on a lot of the above would shade into lying. Inflammatory comments and hiring mostly dems, sure, conceded. Those were mistakes. But Anthropic's policy agenda and Dario's policy-related comments come from sincere beliefs about risks, and moderating on those by deliberately not saying what they *really* believe--on the issues they think are more important than anything else!--seems like it could lead to bad places. Maybe the realpolitik choice is to get comfortable with a little bit of, uh, creatively reframing the truth, and some misleadingness is just what it takes to do politics, but it's a hard bullet to bite if you're afraid of China building a superintelligence that takes over the world. Seems like it merits a bit more of a "Yeah, this sucks and is kind of wrong, but you have to do it because reality isn't leaving you with any better options".
(Also, to be clear, I really deeply appreciate your willingness to criticize your own party. As someone who's more toward the pause-ish end of the spectrum, I get more out of your writing than almost any other source on AI policy. Please keep it up!)
Agree. The politics here come down to Anthropic hiring some former Biden officials, and insisting on some minimal restrictions on model useage. The reaction to all this, as Dean has outlined previously, seems overblown and counter productive, given Anthropic's position among leading AI labs. Here, one wonders whether CAISI was involved in anything resembling pre-clearance of Fable, and whether in fact the assessment of AWS was overblown. Why would we think AWS knows more about the model than Anthropic and is in a better position to determine vulnerabilities? I am left wondering if it was actually a China connection, coming from somewhere behind closed doors, that itself could be overblown of course, that triggered the action, but an doubtful that anyone would be in a position to determine what a Chinese entity may not have gained from access to Fable....
The most alarming detail isn't any single decision. It's the velocity of the swing. The same administration went from "we will not create a licensing regime" to "worldwide export controls on a single model" in a matter of weeks, and the trigger wasn't a change in the technology or a change in the assessed risk. It was a breakdown in a relationship.
That's the structural problem your FDA analogy is really pointing at. When governance runs on relationships rather than institutions, the rules change at the speed of the relationship, not at the speed of the technology. Every company in the field is now operating under a regime that could shift overnight depending on who falls out with whom next. And the damage isn't to any one company. It's to the predictability that every other company, investor, and allied government needs in order to plan at all. The irony is that the stated goal is American dominance in AI, and the one thing that reliably kills a country's lead in a technology race is making the rules unpredictable enough that capital and talent start looking for somewhere more stable to build.
Hopefully more details will be publicly released for people to verify.
To my knowledge: The saddest part of this, was all triggered by someone typing “Fix this code” into Fable and it finding bugs and fixing them.
They _manually_ took that bug fix and reverse engineered it into a 'potential' malicious attack, it would likely go no where or do anything of consequence. The LLM blocked them from automating that process like it was suppose to; so it had to be done by human hands with the prerequisite knowledge. It spooked the Amazon CEO into claiming it was a jailbreak, what a clown.
People are being fearful of A.I. / LLMs being able to find any kind of bugs in code. It looks like it may lead to heavy handed regulation of all LLMs. There are tools that can do all of this on the market already for decades without the use of LLMs to varying degrees they've never been considered a major threat.
I think they're terrified of adversaries using these systems to fix their own code as much they fear it being used to find and create vulnerabilities.
The WH has a few too many idiots at this point based on the actions they're taking. They refuse to learn about technology and refuse to listen to experts that don't agree with their delusional fantasies. Not thinking about the long term consequences of the export controls destroying every research lab in the U.S. the subsequent potential brain drain of the U.S. and everything A.I. related collapsing that could result in a market crash on the level of something no living person has experienced.
Thanks! I’m not suggesting ant should have lobbied for bills to which they have conscientious objection. But in dc, the median bill fails. So if you see a bill you dislike, you needn’t necessarily shout that fact from the rooftops unless you think it has a serious chance at becoming law
If we fail, can we say “if only others had been good, we would have succeeded?” Perhaps. But it does not save us. I disagree with the cowardice and lust for power of the Republican party elites that led us to this position, that they welcomed Donald Trump because the alternative was, gasp, Hillary Clinton. But Anthropic, and the rest of us, must live with the world as it is.
I'm all for realism, not useless displays of virtue. (But is madness ever bound by what is real?) Anyway, it's clear there's a collective action problem here, as aptly demonstrated by Altman. But I'd warn against the way of the lackey, too. Bit by bit, it erodes your moral character and not just in the mirror or in the shadow of power, but in full view of the public.
Of course, there is rarely morality without moralism, without at least a whiff of hypocrisy. It's a big reason why many love the Mad King, too, isn't it? He just does not give a fu**. Yet revulsion is a most natural reaction to boot licking, whether you're wearing one or just watching. That's a losing game if you believe in civic virtue and all that. You convert no enemies and just further demoralize your friends.
If you're playing the morals card, I guess you'll have to welcome the trouble at some point anyway. Put some skin in the game, as they say.
This essay is well written and actually interesting but it still contains the serious blindspot/analytical hole of looking at "the state," "the polity", "democracy" just as abstractions. And it references the founders' fear of the "raw will of the majority" but doesnt seem to want to look too hard at the institutional question beneath that, he looks at mix of representative institutions (good for him), technocratic governance, and "overarching" national frameworks. But there is another path available that never really enters the discussion. If one is concerned about the dangers of simple majoritarianism, the choice is not merely between direct popular impulses and expert administration. There is also the possibility of federated decision-making, distributing authority across states, municipalities, professions, industries, universities, and other institutions so that disagreement is accommodated through policy variability and variation and cooperation instead resolved by the brute force of "raw" technocratic centralized decision making
More importantly, the essay seems to just assumes that AI governance needs to occur through an overarching national framework managed by "raw" technocratic force. But AI is exactly the sort of technology where preferences will vary dramatically across industries, professions, regions, and communities. So I'm perplexed as to why the author never explores federated decision-making with overlapping centers of authority, policy variability, institutional plurality, and the ability to experiment with different governance models simultaneously. That is not only likely to produce better information and better AI deployment outcomes, it is also much closer to what lower-case "d" democracy actually means in a large and diverse society than a single national framework run by a fusion of between central government authorities, frontier firms, the military and then executed by a hierarchical command structures apparatchiks wielding "raw" technocratic force
Independent oversight body with dual reporting responsibilities to executive/legislative branches that examines AI use across departmental silos. GAO/IG for AI. GAIOG
Alas, The author is correct. If Elon Musk or any of the current administration’s preferred vendors released this product, we would never have heard of the “jail break” problem. Amazon’s involvement is an interesting one. Anthropic made choices and we can agree or disagree. The question for America is one of trust. A company that tries to explain risks gets penalized (one on the wrong side of the political spectrum). Where is Andreesen / Horowitz or Peter Thiel standing up for Anthropic? In the end, it is all about money. And guess what, you can be the first trillionaire. But, as Marcus Aurelius wrote, "Alexander the Great and his mule driver both died and the same thing happened to both. They were absorbed alike into the life force of the world, or dissolved alike into atoms.” You can’t have it both ways - or, if you have enough money, maybe you can.
This is a game of chicken with real consequences. Who do you trust? The essay has many good points. But, what is lacking, is what is best for the American people and the world at large. Alignment is not just an AI problem, it is an American political problem.
"Imagine that there were no Food and Drug Administration (FDA), but there remained a large pharmaceutical sector, similar in size and scope to the one the United States enjoys today. In this alternate world, imagine that drugs were not licensed or otherwise formally approved by regulators"
In this world, the cost to bring drugs to market would be dramatically less, meaning the costs of drugs themselves would be less, and alternative therapies, like peptides, would be more widely accepted, helping many. Quality would be enforced by private sector and doctors. That plus more innovation and a lower-cost path to getting drugs on the market will have its own downsides and risks but on balance likely be a safer and healthier America.
But what happens when a drug is widely available and causes great harm? (One thinks of the Opiod over-prescribing scandal). Quality and risk failures in the private sector cause a 'moral panic' and drive the need for Govt regulation. Thalidomide's dangers to babies was used as an example of why we need the FDA (although it came to market with an FDA, not without one).
The issue is that a "safety-first" regime is safer in the short run, but by slowing innovation ends up causing more harm than good in the long run. The FDA has made many mistakes and one can argue that they've done more harm than good on many fronts.
The same is playing out in AI. a kid commits suicide after a ChatGPT convo and the parents sue OpenAI. Is it liable? Lose/lose for AI companies wrt regulation. If they push zero regulation, then they are like tobacco companies; if they go the Anthropic route and overplay the risks, they create the stampede for over-reaction (as happened with Fable) and it sounds suspiciously like a push for 'regulatory capture'.
We will end up with Govt regulation not because it is the best of all options, but because it is the best way for both AI companies and the Govt to CYA when something bad happens. Which it will.
“One hope of mine is that there are ways to design institutions that minimize political interference in technocratic matters, which has been the focus of my writing on private governance and independent verification organizations.”
As a political scientist, I hear you. I wrote about precisely the need to look at voluntary governance models for AI regulation in a short article I published in AI & Society, available here: https://link.springer.com/article/10.1007/s00146-025-02579-1
This is the gist of the problem. According to Trump’s series of AI declarations, there should be no impediments to the race for global AI leadership—no state regulations, no burdensome environmental regulations, etc.
Yet Anthropic is viewed as the problem, an antagonist, for having a different philosophical and political view? That's the issue with federal preemption. Why should the party in power kneecap a leading AI model over ideological differences? Doesn't that conflict with the list of Trump’s directives to be the global AI leader?
I appreciate the thought you’ve put into this analysis and the effort to grapple with the genuinely hard governance problems around frontier AI. That said, I’m struggling with how much of the central narrative is verifiable and how much is inference or framing.
A few examples:
The description of the Anthropic–administration conflict reads as a fairly linear story of “company releases model → jailbreak appears → administration demands takedown → export controls follow.” From the outside, the public record so far is much more fragmentary and ambiguous. Most of what we have are brief government orders, limited company statements, and media accounts that emphasize how little detail has been made available about the underlying “national security” concerns. Leaning too heavily on a single causal story risks overstating what we actually know.
Executive Order 14409 is repeatedly invoked in the piece as if its meaning were unambiguous and clearly at odds with what has happened here. But the order itself is very specific: it creates a voluntary 30‑day pre‑deployment framework for “covered frontier models,” and it explicitly disclaims creating a mandatory licensing or preclearance regime. It does not address the separate national‑security and export‑control authorities that appear to have been used in the Anthropic case. Treating the EO as if it controlled everything the government did here underplays the complexity of the legal toolkit the executive already has.
Likewise, the comparison to the FDA implies a kind of unitary, technocratic gatekeeper that simply doesn’t exist yet for AI. The present situation looks more like overlapping, partially improvised use of cybersecurity, export‑control, and national‑security powers than a coherent “AI FDA.” That doesn’t mean you’re wrong to call for more structured, technocratic institutions—but it does mean we should be careful not to retroactively read a single, clean regulatory logic into what has been, so far, a very ad hoc response.
More broadly, the essay repeatedly characterizes the administration’s motives in strongly psychological and political terms: a desire for “domination” over “disobedient, obstinate, literal nerds,” or actions driven primarily by partisan animus toward Anthropic. Those may turn out to be correct as a matter of political interpretation, but from a strictly evidentiary standpoint they seem to go well beyond what the public record currently supports. A more explicit separation between “here is what we can document” and “here is my political read on why they did it” would make the piece easier to defend and engage with.
For readers who are trying to understand the underlying law and policy, it would also be helpful to see more precise sourcing and clearer boundaries between:
Facts we know (texts of orders, public statements, model shutdowns);
Reasonable inferences from those facts; and
Normative judgments about fairness, proportionality, and political motivation.
The underlying thesis—that we badly need a more coherent, technocratic governance framework that constrains both industry and government—strikes me as important and worth debating. My concern is that the current narrative risks undermining that case by blurring the line between documented events and speculative reconstruction. Tightening that distinction, and dialing back the more speculative attributions of motive, would make the argument both stronger and more persuasive to readers who don’t already share your priors.
Finally, ChatGPT 5.5 is a virtual replica of Mythos yet it was and is released with the same vulnerabilities as Mythos yet the administration did not single out Open Ai in a similar pursuit. of course, MAGA doesn't have the same political agenda against Open AI, but rather, Open AI has agreed to DOW contracts using its AI to illegal surveil American citizens and conduct unmanned drone attacks on targets with no human in the loop.
If we report the facts and the truth, people can make up their own minds. That is the role of a True Journalist.
I notice that mainstream coverage of this is very thin. Not much in the WSJ, etc.
Has anyone explained the Amazon involvement? That also struck me as odd. I assume the government was on a hair trigger looking for a reason to act, but still curious about the claimed direction the feather fell from.
I have some details in another comment here on what happened specifically at Amazon. Official releases on all this have not been made but a few individuals related to this have leaked this information elsewhere if you want to dig around. The sources are from trustworthy long standing members of the cybersec/infosec communities so it's likely accurate.
The Hadfield distinction strikes me as the heart of the essay: fairness questions for politics, technical questions for technocracy. But I wonder how stable that boundary really is in practice. The evaluation of a jailbreak may appear purely technical, yet judgments about acceptable risk, trust, and national security inevitably carry political and normative assumptions. If so, what institutional features prevent private-governance bodies from gradually inheriting the same political pressures they are meant to mediate?
These governance debates tend to fixate on the model layer, who trains what and under which license, because that is where the headlines are. The part that gets less attention and matters more operationally is the deployment layer: what an agent is actually allowed to do once it is running in a real system. You can regulate model training all you like, but the concrete harm surface is an agent with the capability to move money, change infrastructure, or touch data. Governance that cannot reach that layer is mostly symbolic. The enforceable version looks less like licensing and more like access control and audit on actions, which is unglamorous and is where I would put the attention.
After reading this essay, I’m left wondering: What do you think Anthropic should have done with Mythos?
Like, I agree that “tread carefully” is certainly good advice here, but it’s not clear *how* they could’ve tread more carefully. Outside of vibes-y actions like verbally appeasing the Trump admin or hiring a few republicans, I don’t k know what you would have recommended policy-wise. Sit on frontier models until after OpenAI releases a better one? That would cost them the lead in the AI race forever. Run the release by the Trump admin first? What else were Glasswing, the NSA and Pentagon briefings, etc? Add more safety measures? To shut down all GPT-5.5 levels of cyber capabilities *with near-perfect reliability* probably isn’t technically feasible.
The only thing I can think of that might have realistically worked would be waiting for the government to set up their EO cybersecurity clearinghouse, going through it, and then releasing Fable. But I’m not convinced that would’ve been enough. There’s a high chance that the jailbreak would’ve gone unnoticed (I’m not sure the clearinghouse would be Amazon-level right off the bat), and a high chance that the admin would do their inspection, approve it, and then jump on Anthropic anyway as soon as someone finds a partial jailbreak—because they wanted to.
I agree that Dario et al need to take a few levels in realpolitik, but the amount of blame directed at Anthropic feels a bit like “What was she wearing?”
well, look. Anthropic's CEO made inflammatory comments about the President on his personal social media account. He hired many of the senior architects of Biden's AI policy. His company lobbied against the moratorium, and he wrote an oped about it, and this was a priority of some in the admin. His company also lobbied for the GAIN AI Act, a bill the admin did not like, and similarly, Dario wrote an essay about the importance of export controls. Dario made statements about AI and labor that the administration perceived as alarmist and outside his wheelhouse. None of these things advanced the strategic interests of Anthropic, near as I can tell, and all of them antagonized the administration. Any one of them, maybe, is fine, but the combination of them is a pattern of behavior that causes the administration, not without good reason, to see Anthropic as an enemy. And this pattern of behavior also burned political capital that would be coming in handy around the time of the DoW dispute. Ultimately the government is abusing its power, and I think I have been clearer than almost any other Republican about that fact. But Anthropic is behaving as though its rights of free speech also confer a right to infinite political capital. The two are not the same.
Thanks for the reply!
Fair point about the broader history of Anthropic. It does look a bit different when you put everything together.
On the flip side, though, it feels like conceding on a lot of the above would shade into lying. Inflammatory comments and hiring mostly dems, sure, conceded. Those were mistakes. But Anthropic's policy agenda and Dario's policy-related comments come from sincere beliefs about risks, and moderating on those by deliberately not saying what they *really* believe--on the issues they think are more important than anything else!--seems like it could lead to bad places. Maybe the realpolitik choice is to get comfortable with a little bit of, uh, creatively reframing the truth, and some misleadingness is just what it takes to do politics, but it's a hard bullet to bite if you're afraid of China building a superintelligence that takes over the world. Seems like it merits a bit more of a "Yeah, this sucks and is kind of wrong, but you have to do it because reality isn't leaving you with any better options".
(Also, to be clear, I really deeply appreciate your willingness to criticize your own party. As someone who's more toward the pause-ish end of the spectrum, I get more out of your writing than almost any other source on AI policy. Please keep it up!)
Agree. The politics here come down to Anthropic hiring some former Biden officials, and insisting on some minimal restrictions on model useage. The reaction to all this, as Dean has outlined previously, seems overblown and counter productive, given Anthropic's position among leading AI labs. Here, one wonders whether CAISI was involved in anything resembling pre-clearance of Fable, and whether in fact the assessment of AWS was overblown. Why would we think AWS knows more about the model than Anthropic and is in a better position to determine vulnerabilities? I am left wondering if it was actually a China connection, coming from somewhere behind closed doors, that itself could be overblown of course, that triggered the action, but an doubtful that anyone would be in a position to determine what a Chinese entity may not have gained from access to Fable....
Correct. The author doesn't realise anthropic did work with the government and they didn't object to the release.
The most alarming detail isn't any single decision. It's the velocity of the swing. The same administration went from "we will not create a licensing regime" to "worldwide export controls on a single model" in a matter of weeks, and the trigger wasn't a change in the technology or a change in the assessed risk. It was a breakdown in a relationship.
That's the structural problem your FDA analogy is really pointing at. When governance runs on relationships rather than institutions, the rules change at the speed of the relationship, not at the speed of the technology. Every company in the field is now operating under a regime that could shift overnight depending on who falls out with whom next. And the damage isn't to any one company. It's to the predictability that every other company, investor, and allied government needs in order to plan at all. The irony is that the stated goal is American dominance in AI, and the one thing that reliably kills a country's lead in a technology race is making the rules unpredictable enough that capital and talent start looking for somewhere more stable to build.
Bot block?
Hopefully more details will be publicly released for people to verify.
To my knowledge: The saddest part of this, was all triggered by someone typing “Fix this code” into Fable and it finding bugs and fixing them.
They _manually_ took that bug fix and reverse engineered it into a 'potential' malicious attack, it would likely go no where or do anything of consequence. The LLM blocked them from automating that process like it was suppose to; so it had to be done by human hands with the prerequisite knowledge. It spooked the Amazon CEO into claiming it was a jailbreak, what a clown.
People are being fearful of A.I. / LLMs being able to find any kind of bugs in code. It looks like it may lead to heavy handed regulation of all LLMs. There are tools that can do all of this on the market already for decades without the use of LLMs to varying degrees they've never been considered a major threat.
I think they're terrified of adversaries using these systems to fix their own code as much they fear it being used to find and create vulnerabilities.
The WH has a few too many idiots at this point based on the actions they're taking. They refuse to learn about technology and refuse to listen to experts that don't agree with their delusional fantasies. Not thinking about the long term consequences of the export controls destroying every research lab in the U.S. the subsequent potential brain drain of the U.S. and everything A.I. related collapsing that could result in a market crash on the level of something no living person has experienced.
edit: Fixed prompt, put it in quotes.
Thanks! I’m not suggesting ant should have lobbied for bills to which they have conscientious objection. But in dc, the median bill fails. So if you see a bill you dislike, you needn’t necessarily shout that fact from the rooftops unless you think it has a serious chance at becoming law
The King is mad, and his court a council of venomous toadies. Yet you scold the Engineer for failing to read the room?
If we fail, can we say “if only others had been good, we would have succeeded?” Perhaps. But it does not save us. I disagree with the cowardice and lust for power of the Republican party elites that led us to this position, that they welcomed Donald Trump because the alternative was, gasp, Hillary Clinton. But Anthropic, and the rest of us, must live with the world as it is.
I'm all for realism, not useless displays of virtue. (But is madness ever bound by what is real?) Anyway, it's clear there's a collective action problem here, as aptly demonstrated by Altman. But I'd warn against the way of the lackey, too. Bit by bit, it erodes your moral character and not just in the mirror or in the shadow of power, but in full view of the public.
Of course, there is rarely morality without moralism, without at least a whiff of hypocrisy. It's a big reason why many love the Mad King, too, isn't it? He just does not give a fu**. Yet revulsion is a most natural reaction to boot licking, whether you're wearing one or just watching. That's a losing game if you believe in civic virtue and all that. You convert no enemies and just further demoralize your friends.
If you're playing the morals card, I guess you'll have to welcome the trouble at some point anyway. Put some skin in the game, as they say.
This essay is well written and actually interesting but it still contains the serious blindspot/analytical hole of looking at "the state," "the polity", "democracy" just as abstractions. And it references the founders' fear of the "raw will of the majority" but doesnt seem to want to look too hard at the institutional question beneath that, he looks at mix of representative institutions (good for him), technocratic governance, and "overarching" national frameworks. But there is another path available that never really enters the discussion. If one is concerned about the dangers of simple majoritarianism, the choice is not merely between direct popular impulses and expert administration. There is also the possibility of federated decision-making, distributing authority across states, municipalities, professions, industries, universities, and other institutions so that disagreement is accommodated through policy variability and variation and cooperation instead resolved by the brute force of "raw" technocratic centralized decision making
More importantly, the essay seems to just assumes that AI governance needs to occur through an overarching national framework managed by "raw" technocratic force. But AI is exactly the sort of technology where preferences will vary dramatically across industries, professions, regions, and communities. So I'm perplexed as to why the author never explores federated decision-making with overlapping centers of authority, policy variability, institutional plurality, and the ability to experiment with different governance models simultaneously. That is not only likely to produce better information and better AI deployment outcomes, it is also much closer to what lower-case "d" democracy actually means in a large and diverse society than a single national framework run by a fusion of between central government authorities, frontier firms, the military and then executed by a hierarchical command structures apparatchiks wielding "raw" technocratic force
Independent oversight body with dual reporting responsibilities to executive/legislative branches that examines AI use across departmental silos. GAO/IG for AI. GAIOG
Alas, The author is correct. If Elon Musk or any of the current administration’s preferred vendors released this product, we would never have heard of the “jail break” problem. Amazon’s involvement is an interesting one. Anthropic made choices and we can agree or disagree. The question for America is one of trust. A company that tries to explain risks gets penalized (one on the wrong side of the political spectrum). Where is Andreesen / Horowitz or Peter Thiel standing up for Anthropic? In the end, it is all about money. And guess what, you can be the first trillionaire. But, as Marcus Aurelius wrote, "Alexander the Great and his mule driver both died and the same thing happened to both. They were absorbed alike into the life force of the world, or dissolved alike into atoms.” You can’t have it both ways - or, if you have enough money, maybe you can.
This is a game of chicken with real consequences. Who do you trust? The essay has many good points. But, what is lacking, is what is best for the American people and the world at large. Alignment is not just an AI problem, it is an American political problem.
"Imagine that there were no Food and Drug Administration (FDA), but there remained a large pharmaceutical sector, similar in size and scope to the one the United States enjoys today. In this alternate world, imagine that drugs were not licensed or otherwise formally approved by regulators"
In this world, the cost to bring drugs to market would be dramatically less, meaning the costs of drugs themselves would be less, and alternative therapies, like peptides, would be more widely accepted, helping many. Quality would be enforced by private sector and doctors. That plus more innovation and a lower-cost path to getting drugs on the market will have its own downsides and risks but on balance likely be a safer and healthier America.
But what happens when a drug is widely available and causes great harm? (One thinks of the Opiod over-prescribing scandal). Quality and risk failures in the private sector cause a 'moral panic' and drive the need for Govt regulation. Thalidomide's dangers to babies was used as an example of why we need the FDA (although it came to market with an FDA, not without one).
The issue is that a "safety-first" regime is safer in the short run, but by slowing innovation ends up causing more harm than good in the long run. The FDA has made many mistakes and one can argue that they've done more harm than good on many fronts.
The same is playing out in AI. a kid commits suicide after a ChatGPT convo and the parents sue OpenAI. Is it liable? Lose/lose for AI companies wrt regulation. If they push zero regulation, then they are like tobacco companies; if they go the Anthropic route and overplay the risks, they create the stampede for over-reaction (as happened with Fable) and it sounds suspiciously like a push for 'regulatory capture'.
We will end up with Govt regulation not because it is the best of all options, but because it is the best way for both AI companies and the Govt to CYA when something bad happens. Which it will.
“One hope of mine is that there are ways to design institutions that minimize political interference in technocratic matters, which has been the focus of my writing on private governance and independent verification organizations.”
As a political scientist, I hear you. I wrote about precisely the need to look at voluntary governance models for AI regulation in a short article I published in AI & Society, available here: https://link.springer.com/article/10.1007/s00146-025-02579-1
This is the gist of the problem. According to Trump’s series of AI declarations, there should be no impediments to the race for global AI leadership—no state regulations, no burdensome environmental regulations, etc.
Yet Anthropic is viewed as the problem, an antagonist, for having a different philosophical and political view? That's the issue with federal preemption. Why should the party in power kneecap a leading AI model over ideological differences? Doesn't that conflict with the list of Trump’s directives to be the global AI leader?
Love it
I appreciate the thought you’ve put into this analysis and the effort to grapple with the genuinely hard governance problems around frontier AI. That said, I’m struggling with how much of the central narrative is verifiable and how much is inference or framing.
A few examples:
The description of the Anthropic–administration conflict reads as a fairly linear story of “company releases model → jailbreak appears → administration demands takedown → export controls follow.” From the outside, the public record so far is much more fragmentary and ambiguous. Most of what we have are brief government orders, limited company statements, and media accounts that emphasize how little detail has been made available about the underlying “national security” concerns. Leaning too heavily on a single causal story risks overstating what we actually know.
Executive Order 14409 is repeatedly invoked in the piece as if its meaning were unambiguous and clearly at odds with what has happened here. But the order itself is very specific: it creates a voluntary 30‑day pre‑deployment framework for “covered frontier models,” and it explicitly disclaims creating a mandatory licensing or preclearance regime. It does not address the separate national‑security and export‑control authorities that appear to have been used in the Anthropic case. Treating the EO as if it controlled everything the government did here underplays the complexity of the legal toolkit the executive already has.
Likewise, the comparison to the FDA implies a kind of unitary, technocratic gatekeeper that simply doesn’t exist yet for AI. The present situation looks more like overlapping, partially improvised use of cybersecurity, export‑control, and national‑security powers than a coherent “AI FDA.” That doesn’t mean you’re wrong to call for more structured, technocratic institutions—but it does mean we should be careful not to retroactively read a single, clean regulatory logic into what has been, so far, a very ad hoc response.
More broadly, the essay repeatedly characterizes the administration’s motives in strongly psychological and political terms: a desire for “domination” over “disobedient, obstinate, literal nerds,” or actions driven primarily by partisan animus toward Anthropic. Those may turn out to be correct as a matter of political interpretation, but from a strictly evidentiary standpoint they seem to go well beyond what the public record currently supports. A more explicit separation between “here is what we can document” and “here is my political read on why they did it” would make the piece easier to defend and engage with.
For readers who are trying to understand the underlying law and policy, it would also be helpful to see more precise sourcing and clearer boundaries between:
Facts we know (texts of orders, public statements, model shutdowns);
Reasonable inferences from those facts; and
Normative judgments about fairness, proportionality, and political motivation.
The underlying thesis—that we badly need a more coherent, technocratic governance framework that constrains both industry and government—strikes me as important and worth debating. My concern is that the current narrative risks undermining that case by blurring the line between documented events and speculative reconstruction. Tightening that distinction, and dialing back the more speculative attributions of motive, would make the argument both stronger and more persuasive to readers who don’t already share your priors.
Finally, ChatGPT 5.5 is a virtual replica of Mythos yet it was and is released with the same vulnerabilities as Mythos yet the administration did not single out Open Ai in a similar pursuit. of course, MAGA doesn't have the same political agenda against Open AI, but rather, Open AI has agreed to DOW contracts using its AI to illegal surveil American citizens and conduct unmanned drone attacks on targets with no human in the loop.
If we report the facts and the truth, people can make up their own minds. That is the role of a True Journalist.
I notice that mainstream coverage of this is very thin. Not much in the WSJ, etc.
Has anyone explained the Amazon involvement? That also struck me as odd. I assume the government was on a hair trigger looking for a reason to act, but still curious about the claimed direction the feather fell from.
I have some details in another comment here on what happened specifically at Amazon. Official releases on all this have not been made but a few individuals related to this have leaked this information elsewhere if you want to dig around. The sources are from trustworthy long standing members of the cybersec/infosec communities so it's likely accurate.
The Hadfield distinction strikes me as the heart of the essay: fairness questions for politics, technical questions for technocracy. But I wonder how stable that boundary really is in practice. The evaluation of a jailbreak may appear purely technical, yet judgments about acceptable risk, trust, and national security inevitably carry political and normative assumptions. If so, what institutional features prevent private-governance bodies from gradually inheriting the same political pressures they are meant to mediate?
These governance debates tend to fixate on the model layer, who trains what and under which license, because that is where the headlines are. The part that gets less attention and matters more operationally is the deployment layer: what an agent is actually allowed to do once it is running in a real system. You can regulate model training all you like, but the concrete harm surface is an agent with the capability to move money, change infrastructure, or touch data. Governance that cannot reach that layer is mostly symbolic. The enforceable version looks less like licensing and more like access control and audit on actions, which is unglamorous and is where I would put the attention.