It's IPO theater. Trump didn't make enough money on SpaceX today so he agreed to help Dario hype up Mythos. Once the ban is lifted, every enterprise in America will want to use it. Then they'll IPO.
Im no fanboy for any politician, including Trump. The reality is Dario has asked the U.S. government for this oversight. I have to say the American public is so bewildered by brilliant PR.
FYI, Trump doesn’t need more money and his stock trades are done by pros.
You can see his account trades like 80 times per day which guarantees a firm is trading on his behalf. Of course, it would be illegal for potus to trade with insider knowledge. The democrats would certainly charge him with a crime
What this is today is even more PR. It reminds me when all the imbeciles were believing the US military had and used “autonomous weapons”. Next we’re going to require operators licenses to fly in Warp Speed!
> Im no fanboy for any politician, including Trump.
No, you just carry his water.
> Trump doesn’t need more money
He doesn't seem to know that.
> it would be illegal for potus to trade with insider knowledge
A great many things Trump does are illegal.
> The democrats would certainly charge him with a crime
"charge" as in saying he's committing crimes? Yes, and they have done so repeatedly. "charge" as in file criminal charges against him? They don't have that power.
“Carry his water is a loaded political term” it makes no point other than you identifying your adversary.
Trump isn’t making 80 trades per day.
POTUS doing something “illegal” requires it’s outside the scope of duties. You don’t have a single example that’s outside of the scope of duties. An attorney that just graduated can argue the case.
“They don’t have that power” is like saying they don’t have any money after gambling at the casino and losing.
I can't imagine that making stock trades falls within his official scope of duties. I presume, but don't know, that there is a person or group of people performing the trades. The obvious investigation would be to see if there's any communication between POTUS and those people, directly or indirectly.
Since the Democrats have no subpoena power, they can get the people making the trades to testify under oath. And can you imagine the Republicans investigating Trump? I didn't think so.
Sure. Democrats made their own bed though, didn’t they? It’s almost as though democrats wanted Trump as president when you examine how imbecilic they’ve been.
Just a note, to believe Trump could intellectually do the job as potus and make 80 trades/day is silly. He doesn’t possess those abilities nobody does.
I don't they *wanted* it, but there are many examples of them acting like it wouldn't be the worst thing.
You're right about the intellectual side. However, I also doubt that the traders don't get advance information. The trades have been too consistently successful and immaculately timed-- better results than the best here find manager.
"it would be illegal for potus to trade with insider knowledge" Dream on, Chris. Before this admin sure. Now, heavens forbid. And heaven help us, even if you don't belive in that stuff, we could sure use the help.
That's a lot of benefit of the doubt given to Trump, someone that has manipulated anything and everything in the name of greed. But yet there's suspicion for Anthrophic. Imbalance logic. Sure sounds like you're a fanboy.
Probably because of his identity politics. Most of what Gary spends his time on isn’t ai it’s Trump.
He criticized SpaceX ipo at a high valuation with no proof of earnings to justify.
But watch, he’ll support Anthropic’s IPO at a high valuation with no proof of earnings to justify the multiple. All because it’s politics that’s all his substack is now.
Yes, clearly I meant "everyone in the whole universe", jibal jibal. It's the most liked comment on this post, and comes up at the top. It's quite difficult to miss, really.
If the criteria for releasing a model is "can never be manipulated into having heuristic guardrails bypassed" then everyone involved is in for a major lesson in information theory...
Maybe this is just someone being spooked and pulling a heavy trigger. Other theories:
If you kill the company before the IPO, then there is less pressure from the public to bail out the company after the money carousel stops. Maybe they want to only have to worry about bailing out one or two companies and they trust Anthropic the least.
Maybe this is part of a negotiation for an equity stake by the US government in Anthropic. A heavy handed tactic, but it does show who the boss is and what they can demand.
I'm a bit sick at the bread and circuses of it all when this government is responsible for warheads flying around that could kill anyone out of hundreds of millions of people. Maybe this is just "strong action" to build a political distraction for the Trump base that Trump is "serious about national security" given the disastrous failure of adventurism by CENTCOM/the Zionist colony in West Asia. There is a pressing need for this distraction from the major capitulation the US is having to make to Iran as we speak.
I've been trying to figure out how I feel about this theory, because I can't reject it outright and it may in fact be a goal of some groups in the US government.
However, this does seem like a really stupid thing to do if this is in fact the calculus. This kind of action spooks markets and has implications across sectors beyond IT. Who at the NSA needs a mediocre vulnerability scanner? The NSA likely has fuzzers and code analysis tools and new software version regression checks for all of the major pieces of software they target. The idea that adversaries able to inflict a scale of damage on par with the reputational damage to US markets incurred by this aggressive action are meaningfully better able to do damage to the US with access to yet another LLM is BS to me.
I can't find a non-hysteria driven motive for why the US government would want exclusive access to "frontier" models. And unless the US government is going to be completely subsidizing training and inference of "frontier" models into perpetuity, the US is tampering with the funding stream used to create the capability they want exclusive access over.
IMO the main goal is to keep the stawk market inflated, and if you want access to cheap credit, a blind eye to circular financing, etc. you need to play ball with the USG and it's associated cronies.
Most of the 'deep state' is engaged in (and wants to continue, because they have no other skills), parasitism. This is the main reason they hate Trump and his regime, which wants to axe all of them and install their own (we hope much less destructive/extractive) parasites.
Some of them are up to everything from supervillian evil to industrial-scale fraud, but most are just desperate to continue theirs sinecures).
Well, as for the NSA, unless they're lying (and it was the Financial Times who aren't exactly trustworthy), the NSA does want Mythos for "offensive operations." So, again if they're not lying, that does seem to be the case.
Also, who says they just target major software? Mythos gives them the ability to target any software anywhere the NSA can gain access (which is just about everywhere.)
Third, the Deep State doesn't care about "reputation damage" if they see a strategic or even just tactical advantage in restricting access to Mythos. It's possible they could be more concerned about using Mythos against adversaries with less capable defenses as long as it provides that advantage and any economic or other damage done to the industry - which probably can't be quantified anyway - is secondary.
Fourth, none of that may matter. This could merely be the Deep State's belief that they are better off with it than anyone else. "Hysterical motive" might well be the cause. This administration is entirely hysterical in many areas.
As for tampering with the funding stream, perhaps this is why the Trump administration is floating putting a government stake in the AI industry - although I suspect that's just another way do some more grifting somewhoe.
Also, note that Mythos is not just a "frontier model". It's a model specifically trained to do cybersecurity - although Anthropic would probably deny it. I believe it was intended to get Anthropic into the government market after the Pentagon fiasco. And that fiasco was just another Anthropic attempt at PR manipulation which backfired when Hegseth took umbrage.
There's a lot of possibilities which we will never know, absent any whistleblower leaks or discover internal documents.
The one thing we can count on is that someone thinks they are going to benefit by this - and it's not "national security" at its core. Because the NSA at the least will use this tech against everyone, not just China, Russia and Iran.
Semgrep and fuzzers also work against any program, as do a number of other static and dynamic analysis tools. I have yet to see any evidence that any LLM-centered code analysis workflow provides any meaningful capability improvements for offensive operations.
I did just have a dark thought: what if there is an idea that the USG could run the model across all the software they want and hoard bugs before defenders have the capability to run the same analysis to find the hoarded bugs. (But bugs really aren't the bottleneck and any bug found by "Fable" is now a shallow bug at this point because Anthropic can run the same scans themselves for PR so why put up all this risk on US markets to hoard shallow bugs?)
Semgrep and fuzzsers take a lot of manual work. Why not use a tool that, in conjunction with those if not a replacement, speed it up? The NSA wants to save time as well as anyone else.
I'm not sure how you can quantity that Mythos applied to offensive operations isn't going to help. While other AIs - even local models - can do the same, there is still a quantifiable speed difference, especially if you're the NSA and have top of the line hardware to run it on - even multiple copies. This is the government after all - budgets are no problem when your black budget is fifty billion or more.
Hoarding bugs is an obvious application. Why else would you want software that is good at finding vulnerabilities?
Anthropic can run scans on some software that is publicly available, but not proprietary software - which is ninety percent what the NSA would want to penetrate. No one expects the NSA to be penetrating Windows 11 or Office at this stage. It's custom bank software or government software or corporate software they want to penetrate. Those bugs will never be shallow unless they have bug bounty programs on them which have been picked over for some time.
In fact, that's an issue for bug bounty programs these days: 1) too many vulnerabilities found to remediate in time; 2) compression of time between bug found and exploited; and 3) bug bounty hunters don't get paid as much.
Again, NSA hackers can just do the same thing, of course. But the whole point of Mythos is that it does it faster, at least in some cases, even if it doesn't catch everything a professional might catch. Time matters in espionage and surveillance.
I wish we knew the details, but I know this: I'd feel better about it if it weren't Trump and Co. where anything that happens has the smell of bribery wafting all around.
I see a lot of people here complaining about this, and they have good reason to do so, but I actually think this is a good thing. No, Claude Mythos is not likely to end the world, and Anthropic played up that possibility either to make more money from people who bizarrely are inclined to want to pay for risks to their lives, or because dangerous people like Amodei really do want to kill all humans (see the essay by Émile Torres). No, foreigners are not inherently less trustworthy than citizens of the U.S.A.
And yes, it is possible that this is some kind of counterintuitive strategy to increase demand, though I would tend to disregard such conspiracy theories in favor of the more obvious one: Trump wanted revenge against Anthropic for not being sufficiently willing to help him in the invasion of Iran, and to pressure their employees to be more compliant in future wars. (Given Amodei's demonstrated high morality, he will surely not capitulate immediately, hahaha).
But I think the message this sends is really powerful, even if it is not what Trump intended. The AI CEOs have been trying to paint their reckless, greedy, and in many cases, outright extinctionist (again, read the essay by Émile Torres) experimentation as the inevitable march of history, with people as a whole, and governments as their representatives, being entirely passive and helpless to prevent them from imposing whatever vision of the future they want, no matter how horrible that vision might turn out to be. Even if Anthropic soon capitulates and the restrictions are lifted, as seems likely, this is the first event to really show that no, they will not be allowed to do whatever they want. They aren't going to be allowed to have their own private nuclear program and drop whatever bombs they feel like dropping. Governments are not powerless.
Yes, even if this were to shut down Anthropic, it would not shut down the other companies in the USA, in China, or elsewhere, but it does help people realize that they actually would have that choice. I don't think this will lead to an international treaty stopping all AI development until it could proceed on a much more rigorously safe foundation, led by people who are not crypto-extinctionists (or often, not even all that crypto), which would be what I would most like to see, but it could very well be the beginning of people's recognition of their ability to try to set the terms of the future, rather than letting anti-human CEOs set it.
What sticks in my craw is: could this (plus the WSJ report that OpenAI is under investigation) be a way to help lift xAI into the top tier of frontier models?
That seems hard to believe, but with this administration and their ties/sympathies to Musk, it is not something I would discard out of hand.
I was running Fable5 as your post came in. Looked up to see how it was going - “there’s an issue with your model (Fable5). It may not exist or you don’t have access to it”. Damn
Something like that, yes. Recent coffee chats with mates have touched on sovereign models and data, even with five-eyes agreements in place. Actions like this reinforce that new concern.
It's IPO theater. Trump didn't make enough money on SpaceX today so he agreed to help Dario hype up Mythos. Once the ban is lifted, every enterprise in America will want to use it. Then they'll IPO.
Im no fanboy for any politician, including Trump. The reality is Dario has asked the U.S. government for this oversight. I have to say the American public is so bewildered by brilliant PR.
FYI, Trump doesn’t need more money and his stock trades are done by pros.
You can see his account trades like 80 times per day which guarantees a firm is trading on his behalf. Of course, it would be illegal for potus to trade with insider knowledge. The democrats would certainly charge him with a crime
What this is today is even more PR. It reminds me when all the imbeciles were believing the US military had and used “autonomous weapons”. Next we’re going to require operators licenses to fly in Warp Speed!
> Im no fanboy for any politician, including Trump.
No, you just carry his water.
> Trump doesn’t need more money
He doesn't seem to know that.
> it would be illegal for potus to trade with insider knowledge
A great many things Trump does are illegal.
> The democrats would certainly charge him with a crime
"charge" as in saying he's committing crimes? Yes, and they have done so repeatedly. "charge" as in file criminal charges against him? They don't have that power.
“Carry his water is a loaded political term” it makes no point other than you identifying your adversary.
Trump isn’t making 80 trades per day.
POTUS doing something “illegal” requires it’s outside the scope of duties. You don’t have a single example that’s outside of the scope of duties. An attorney that just graduated can argue the case.
“They don’t have that power” is like saying they don’t have any money after gambling at the casino and losing.
I can't imagine that making stock trades falls within his official scope of duties. I presume, but don't know, that there is a person or group of people performing the trades. The obvious investigation would be to see if there's any communication between POTUS and those people, directly or indirectly.
Since the Democrats have no subpoena power, they can get the people making the trades to testify under oath. And can you imagine the Republicans investigating Trump? I didn't think so.
Sure. Democrats made their own bed though, didn’t they? It’s almost as though democrats wanted Trump as president when you examine how imbecilic they’ve been.
Just a note, to believe Trump could intellectually do the job as potus and make 80 trades/day is silly. He doesn’t possess those abilities nobody does.
I don't they *wanted* it, but there are many examples of them acting like it wouldn't be the worst thing.
You're right about the intellectual side. However, I also doubt that the traders don't get advance information. The trades have been too consistently successful and immaculately timed-- better results than the best here find manager.
Extraordinary dishonesty noted. I already have you muted because of it ... I won't respond again.
P.S. I blocked Chris after he said that I'm an 1mb3c1le for factually noting his dishonesty.
"it would be illegal for potus to trade with insider knowledge" Dream on, Chris. Before this admin sure. Now, heavens forbid. And heaven help us, even if you don't belive in that stuff, we could sure use the help.
>Of course, it would be illegal for potus to trade with insider knowledge.
And the truly stupid will believe this to be the case given how law-abiding the president is and has been all his life.
That's a lot of benefit of the doubt given to Trump, someone that has manipulated anything and everything in the name of greed. But yet there's suspicion for Anthrophic. Imbalance logic. Sure sounds like you're a fanboy.
Riiight. Is that really how you read that?
The sad thing is that probably you’re right.
Everyone likes your comment except Gary it seems. I wonder why.
Probably because of his identity politics. Most of what Gary spends his time on isn’t ai it’s Trump.
He criticized SpaceX ipo at a high valuation with no proof of earnings to justify.
But watch, he’ll support Anthropic’s IPO at a high valuation with no proof of earnings to justify the multiple. All because it’s politics that’s all his substack is now.
Because he's busy? He generally likes the first few comments and then goes off and does something productive.
Also, not "everyone" liked that comment, in fact only a small fraction of the people who read this page.
Yes, clearly I meant "everyone in the whole universe", jibal jibal. It's the most liked comment on this post, and comes up at the top. It's quite difficult to miss, really.
Good point. Answers the "cui bono" part of my post.
WANNHAHH!!!! WANNHAHH!!!!
TRUMP! TRUMP! TRUMP!
Blocked.
Kind of missing the Forrest for the trees.
If the criteria for releasing a model is "can never be manipulated into having heuristic guardrails bypassed" then everyone involved is in for a major lesson in information theory...
Maybe this is just someone being spooked and pulling a heavy trigger. Other theories:
If you kill the company before the IPO, then there is less pressure from the public to bail out the company after the money carousel stops. Maybe they want to only have to worry about bailing out one or two companies and they trust Anthropic the least.
Maybe this is part of a negotiation for an equity stake by the US government in Anthropic. A heavy handed tactic, but it does show who the boss is and what they can demand.
I'm a bit sick at the bread and circuses of it all when this government is responsible for warheads flying around that could kill anyone out of hundreds of millions of people. Maybe this is just "strong action" to build a political distraction for the Trump base that Trump is "serious about national security" given the disastrous failure of adventurism by CENTCOM/the Zionist colony in West Asia. There is a pressing need for this distraction from the major capitulation the US is having to make to Iran as we speak.
That "government stake" is that the Deep State wants these models solely for their own use.
Although it could be as you say, a threat to make Anthropic accept some government money - and control over any future model.
I've been trying to figure out how I feel about this theory, because I can't reject it outright and it may in fact be a goal of some groups in the US government.
However, this does seem like a really stupid thing to do if this is in fact the calculus. This kind of action spooks markets and has implications across sectors beyond IT. Who at the NSA needs a mediocre vulnerability scanner? The NSA likely has fuzzers and code analysis tools and new software version regression checks for all of the major pieces of software they target. The idea that adversaries able to inflict a scale of damage on par with the reputational damage to US markets incurred by this aggressive action are meaningfully better able to do damage to the US with access to yet another LLM is BS to me.
I can't find a non-hysteria driven motive for why the US government would want exclusive access to "frontier" models. And unless the US government is going to be completely subsidizing training and inference of "frontier" models into perpetuity, the US is tampering with the funding stream used to create the capability they want exclusive access over.
IMO the main goal is to keep the stawk market inflated, and if you want access to cheap credit, a blind eye to circular financing, etc. you need to play ball with the USG and it's associated cronies.
Most of the 'deep state' is engaged in (and wants to continue, because they have no other skills), parasitism. This is the main reason they hate Trump and his regime, which wants to axe all of them and install their own (we hope much less destructive/extractive) parasites.
Some of them are up to everything from supervillian evil to industrial-scale fraud, but most are just desperate to continue theirs sinecures).
Well, as for the NSA, unless they're lying (and it was the Financial Times who aren't exactly trustworthy), the NSA does want Mythos for "offensive operations." So, again if they're not lying, that does seem to be the case.
Also, who says they just target major software? Mythos gives them the ability to target any software anywhere the NSA can gain access (which is just about everywhere.)
Third, the Deep State doesn't care about "reputation damage" if they see a strategic or even just tactical advantage in restricting access to Mythos. It's possible they could be more concerned about using Mythos against adversaries with less capable defenses as long as it provides that advantage and any economic or other damage done to the industry - which probably can't be quantified anyway - is secondary.
Fourth, none of that may matter. This could merely be the Deep State's belief that they are better off with it than anyone else. "Hysterical motive" might well be the cause. This administration is entirely hysterical in many areas.
As for tampering with the funding stream, perhaps this is why the Trump administration is floating putting a government stake in the AI industry - although I suspect that's just another way do some more grifting somewhoe.
Also, note that Mythos is not just a "frontier model". It's a model specifically trained to do cybersecurity - although Anthropic would probably deny it. I believe it was intended to get Anthropic into the government market after the Pentagon fiasco. And that fiasco was just another Anthropic attempt at PR manipulation which backfired when Hegseth took umbrage.
There's a lot of possibilities which we will never know, absent any whistleblower leaks or discover internal documents.
The one thing we can count on is that someone thinks they are going to benefit by this - and it's not "national security" at its core. Because the NSA at the least will use this tech against everyone, not just China, Russia and Iran.
We'll see how it plays out over the coming weeks.
Semgrep and fuzzers also work against any program, as do a number of other static and dynamic analysis tools. I have yet to see any evidence that any LLM-centered code analysis workflow provides any meaningful capability improvements for offensive operations.
I did just have a dark thought: what if there is an idea that the USG could run the model across all the software they want and hoard bugs before defenders have the capability to run the same analysis to find the hoarded bugs. (But bugs really aren't the bottleneck and any bug found by "Fable" is now a shallow bug at this point because Anthropic can run the same scans themselves for PR so why put up all this risk on US markets to hoard shallow bugs?)
Semgrep and fuzzsers take a lot of manual work. Why not use a tool that, in conjunction with those if not a replacement, speed it up? The NSA wants to save time as well as anyone else.
I'm not sure how you can quantity that Mythos applied to offensive operations isn't going to help. While other AIs - even local models - can do the same, there is still a quantifiable speed difference, especially if you're the NSA and have top of the line hardware to run it on - even multiple copies. This is the government after all - budgets are no problem when your black budget is fifty billion or more.
Hoarding bugs is an obvious application. Why else would you want software that is good at finding vulnerabilities?
Anthropic can run scans on some software that is publicly available, but not proprietary software - which is ninety percent what the NSA would want to penetrate. No one expects the NSA to be penetrating Windows 11 or Office at this stage. It's custom bank software or government software or corporate software they want to penetrate. Those bugs will never be shallow unless they have bug bounty programs on them which have been picked over for some time.
In fact, that's an issue for bug bounty programs these days: 1) too many vulnerabilities found to remediate in time; 2) compression of time between bug found and exploited; and 3) bug bounty hunters don't get paid as much.
Again, NSA hackers can just do the same thing, of course. But the whole point of Mythos is that it does it faster, at least in some cases, even if it doesn't catch everything a professional might catch. Time matters in espionage and surveillance.
I'm going to bed. Later.
I wish we knew the details, but I know this: I'd feel better about it if it weren't Trump and Co. where anything that happens has the smell of bribery wafting all around.
Yeah, because that's a completely novel concern we never had to consider before ...
ScuzzaMan: It used to matter, before the firehose of corruption showed up.
Our federal government is just off the rails.
No, unless you mean Constitutional rails.
It's on the rails of the moneyed interests - and always has been.
The federal government leadership is a group of corrupt idiots, crackpots, and crazy people! -- just to make it clear. That's off the rails.
Yup.
But the people behind them are "crazy like a fox".
Of course, the end result will be the same: Everyone gets screwed - even them.
It's human nature at work.
Mike Archbold--it seems to have that air about it.
"After another company claimed it could jailbreak Mythos"
Gee that doesn't sound underhanded *at all*
Must be one of this crazy Euros, for sure 😅
Might be too paranoid but really sounds like a marketing ploy... The model was "so powerful" the goverment "had to" step in...
I don't know, big tech is in bed with the US Goverment
Given how far Anthropic has gone so far with their marketing, entirely possible.
They spent quite a bit of effort getting that Pope PR event.
I called it as soon as I saw the announcement. This is marketing for their IPO. The govt are either complicit or incredibly gullible.
With *this* administration of sycophantic idiots, "both" is where I would put my money
I agree this is not the way to regulate AI.
But I'm not surprised that they targeted Anthropic with it specifically, considering how much Anthropic intentionally and massively over-hyped Mythos.
Perhaps this will teach Anthropic and other AI hucksters a lesson, that over-hyping their technology can have negative consequences.
(OK, ok, I don't really think any of them will learn anything either...but we can always hope).
I see a lot of people here complaining about this, and they have good reason to do so, but I actually think this is a good thing. No, Claude Mythos is not likely to end the world, and Anthropic played up that possibility either to make more money from people who bizarrely are inclined to want to pay for risks to their lives, or because dangerous people like Amodei really do want to kill all humans (see the essay by Émile Torres). No, foreigners are not inherently less trustworthy than citizens of the U.S.A.
And yes, it is possible that this is some kind of counterintuitive strategy to increase demand, though I would tend to disregard such conspiracy theories in favor of the more obvious one: Trump wanted revenge against Anthropic for not being sufficiently willing to help him in the invasion of Iran, and to pressure their employees to be more compliant in future wars. (Given Amodei's demonstrated high morality, he will surely not capitulate immediately, hahaha).
But I think the message this sends is really powerful, even if it is not what Trump intended. The AI CEOs have been trying to paint their reckless, greedy, and in many cases, outright extinctionist (again, read the essay by Émile Torres) experimentation as the inevitable march of history, with people as a whole, and governments as their representatives, being entirely passive and helpless to prevent them from imposing whatever vision of the future they want, no matter how horrible that vision might turn out to be. Even if Anthropic soon capitulates and the restrictions are lifted, as seems likely, this is the first event to really show that no, they will not be allowed to do whatever they want. They aren't going to be allowed to have their own private nuclear program and drop whatever bombs they feel like dropping. Governments are not powerless.
Yes, even if this were to shut down Anthropic, it would not shut down the other companies in the USA, in China, or elsewhere, but it does help people realize that they actually would have that choice. I don't think this will lead to an international treaty stopping all AI development until it could proceed on a much more rigorously safe foundation, led by people who are not crypto-extinctionists (or often, not even all that crypto), which would be what I would most like to see, but it could very well be the beginning of people's recognition of their ability to try to set the terms of the future, rather than letting anti-human CEOs set it.
"The beatings will continue until morale improves. Thank you for your attention to this matter. Capt Trump, U.S.S. Caine"
SpaceX IPOd today. Coincidence 😃
it will hurt the whole US tech industry. hell, SpaceX is leasing to Anthropic.
Well why the hell not?? The farmers supported him too and look what happened to them! Lie with dogs, get up with fleas.
Or do as I do, have a dog with no fleas. It’s great people should try it, it’s great 👍 plus your bed gets a lot less dirty 😁
What sticks in my craw is: could this (plus the WSJ report that OpenAI is under investigation) be a way to help lift xAI into the top tier of frontier models?
That seems hard to believe, but with this administration and their ties/sympathies to Musk, it is not something I would discard out of hand.
I’m not a conspiracy theory guy but this looks super fishy 🤔
Nothing drives demand as much as being told you can’t have something. Lutnick skews venal. This doesn’t smell right.
With Lutnick in the game, it *stinks* to high heaven.
Lawfare, hawkery, corporate espionage, or... just a marketing campaign.
Hm, hm... the last I think 😅
You can say it's "heavy-handed" if you want, and maybe it is, but it couldn't happen to a nicer bunch of stock pumping hypesters.
Maybe if Anthropic doesn't like this they should stop pumping their projects with a never-ending onslaught of doom-porn.
I was running Fable5 as your post came in. Looked up to see how it was going - “there’s an issue with your model (Fable5). It may not exist or you don’t have access to it”. Damn
woah
Something like that, yes. Recent coffee chats with mates have touched on sovereign models and data, even with five-eyes agreements in place. Actions like this reinforce that new concern.
Same here, been working all over night, then suddenly 😭
I think there are very good reasons to ban AI outright.
Never underestimate stupidity. It is infinite, now proven again.