I might have posted this essay, "Will Washington and Beijing Nationalize AI Labs?" as its own topic, but it's a useful analysis of the great power competition implicit in how the U.S. and China...
I might have posted this essay, "Will Washington and Beijing Nationalize AI Labs?" as its own topic, but it's a useful analysis of the great power competition implicit in how the U.S. and China are treating AI model governance.
As far as the Trump administration is concerned, I don't know if there's deeper strategy involved than rewarding loyalty and punishing dissent, but Chinese governance in general is a matter of running out the lead and tightening when instability appears.
Key quote:
Anthropic is right that frontier systems may create catastrophic misuse risks and that democratic governments need a serious strategy for securing model weights, compute, and dangerous capabilities. But its public framing too often collapses two different claims into one: that advanced AI requires strong safety governance, and that the United States and its leading labs should a preserve decisive advantage over China.
That fusion is analytically unstable, in my view. As the claim that frontier AI could enable catastrophic cyber, biological, or military effects becomes stronger, so too does the case for U.S. controls on China, and, as a result, for China to conclude that U.S. labs are effectively part of a national-security project. In that world, safety rhetoric can become indistinguishable from the U.S. denying Chinese access to key components as a matter of strategy. The result is not necessarily more global safety; it may be more mutual insecurity, more hedging, and more pressure on both governments to pull frontier labs closer to the state.
The US government Friday lifted its block on Anthropic’s powerful Claude Mythos 5 AI model, allowing the company to release it to more than 100 US institutions, including major companies and government agencies.
The decision, in a letter sent Friday afternoon to Anthropic, is a major de-escalation in the confrontation between the Trump Administration and one of the world’s most valuable private companies. Two weeks ago the administration imposed export controls on Mythos, leading to a shut down of the model and its cousin Fable 5 after warnings from Amazon and other companies that they could be “jailbroken” for malicious purposes.
The letter is silent on Fable 5, a weaker version of Mythos that was briefly the most powerful AI model widely available to consumers. People close to the talks said they are moving toward releasing Fable as well, though that timeline is unclear.
“I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” Commerce Secretary Howard Lutnick wrote to Anthropic’s chief compute officer Tom Brown Friday, citing “significant progress” in the intense, daily talks between the government and the company since the block went into effect.
[...]
Under the new Anthropic arrangement, “a license will no longer be required to export, reexport, or in-country transfer (including deemed exports and reexports) the Claude Mythos 5 Model to entities identified in Annex A to this letter and their foreign national employees, or to Anthropic’s foreign national employees.”
I can't speak for Mythos, but at first blush, Fable seemed pretty comparable to Opus. For me, Fable didn't really distinguish itself until I gave it bigger problems or tasks. With Opus, I'd...
I can't speak for Mythos, but at first blush, Fable seemed pretty comparable to Opus.
For me, Fable didn't really distinguish itself until I gave it bigger problems or tasks. With Opus, I'd typically have to go through several back and forths dialing everything in. But Fable seemed to get what it needed to do a lot more quickly, like it really understood the assignment.
That's what made Fable so unique vs. Opus, in my limited experience. It was almost like describing a task to a competent person who's completely new to a project vs describing it to someone who's been around for a while.
I wonder what the actual story is here? Is this the next step in getting OpenAI and Anthropic bailed out by the government? Did someone realize government software is such garbage they are worried...
I wonder what the actual story is here?
Is this the next step in getting OpenAI and Anthropic bailed out by the government?
Did someone realize government software is such garbage they are worried a LLM could do some damage?
The next Trump grift where he tries to figure out a way to enrich himself by forcing these companies to pay him off, more than they already have been?
OpenAI did in the past. There is no reason to think that Anthropic wouldn't want the same thing. If they can't IPO at the valuation investors need, which increasingly is rumoured to be the case,...
OpenAI did in the past. There is no reason to think that Anthropic wouldn't want the same thing. If they can't IPO at the valuation investors need, which increasingly is rumoured to be the case, then the next thing to do is convince everyone that your model is so amazing and dangerous the government needs to invest money to keep it from "falling in the wrong hands", etc.
The way they exaggerate a talking point about a possible loan guarantee into a “asking for a bailout” shows why they’re not to be trusted. At the same time, I don’t see why someone at OpenAI...
The way they exaggerate a talking point about a possible loan guarantee into a “asking for a bailout” shows why they’re not to be trusted.
At the same time, I don’t see why someone at OpenAI thought that could possibly happen.
Yeah, basically that. Their CFO talked about it, which was a weird thing for the CFO of a "successful" company like OpenAI to do unless they knew they were going to have a hard time raising enough...
Yeah, basically that. Their CFO talked about it, which was a weird thing for the CFO of a "successful" company like OpenAI to do unless they knew they were going to have a hard time raising enough money for the AI furnace.
OpenAI giving the government stock for free would not be a bailout. Unless there's money moving in the other direction. Maybe it's more like a bribe, but the question is what would OpenAI get for it.
OpenAI giving the government stock for free would not be a bailout. Unless there's money moving in the other direction.
Maybe it's more like a bribe, but the question is what would OpenAI get for it.
I think there’s a simpler explanation? People—or at least the executives at many companies involved—believe these models are quite powerful and want to justify government controls so they have...
I think there’s a simpler explanation?
People—or at least the executives at many companies involved—believe these models are quite powerful and want to justify government controls so they have access to them before competitors do. Business as usual; capital building moats with regulatory capture!
The stated goal of having only a limited release of Mythos was to allow defenders to patch security bugs in important software before attackers get a chance to exploit them. I think that still...
The stated goal of having only a limited release of Mythos was to allow defenders to patch security bugs in important software before attackers get a chance to exploit them. I think that still applies?
I suppose you can twist that into “have access before competitors do” but it depends on what kind of competition you mean. We shouldn’t want to be neutral about the arms race between patching security bugs and exploiting them.
My guess is that the thieves (Anthropic and OpenAI) are afraid that they will be stolen from (via distillation). By releasing to a small set of true users they can build up their distillation...
My guess is that the thieves (Anthropic and OpenAI) are afraid that they will be stolen from (via distillation). By releasing to a small set of true users they can build up their distillation defenses by getting a clear picture of what an actual usage distribution looks like vs a GA release that would definitely contain distillation traffic.
I might have posted this essay, "Will Washington and Beijing Nationalize AI Labs?" as its own topic, but it's a useful analysis of the great power competition implicit in how the U.S. and China are treating AI model governance.
As far as the Trump administration is concerned, I don't know if there's deeper strategy involved than rewarding loyalty and punishing dissent, but Chinese governance in general is a matter of running out the lead and tightening when instability appears.
Key quote:
From the article:
[...]
The real question is... Does Mythos hold up to the hype or is this bullshit?
I can't speak for Mythos, but at first blush, Fable seemed pretty comparable to Opus.
For me, Fable didn't really distinguish itself until I gave it bigger problems or tasks. With Opus, I'd typically have to go through several back and forths dialing everything in. But Fable seemed to get what it needed to do a lot more quickly, like it really understood the assignment.
That's what made Fable so unique vs. Opus, in my limited experience. It was almost like describing a task to a competent person who's completely new to a project vs describing it to someone who's been around for a while.
It depends what hype you were listening to. It's supposed to be pretty good.
I wonder what the actual story is here?
Is this the next step in getting OpenAI and Anthropic bailed out by the government?
Did someone realize government software is such garbage they are worried a LLM could do some damage?
The next Trump grift where he tries to figure out a way to enrich himself by forcing these companies to pay him off, more than they already have been?
Nobody's asking for a bailout. Why bring that up?
OpenAI did in the past. There is no reason to think that Anthropic wouldn't want the same thing. If they can't IPO at the valuation investors need, which increasingly is rumoured to be the case, then the next thing to do is convince everyone that your model is so amazing and dangerous the government needs to invest money to keep it from "falling in the wrong hands", etc.
When did OpenAI ask for a bailout?
I don't know if this is what they are talking about, but prospect.org did report in November 2025 that OpenAI was angling for a government bailout.
The way they exaggerate a talking point about a possible loan guarantee into a “asking for a bailout” shows why they’re not to be trusted.
At the same time, I don’t see why someone at OpenAI thought that could possibly happen.
Yeah, basically that. Their CFO talked about it, which was a weird thing for the CFO of a "successful" company like OpenAI to do unless they knew they were going to have a hard time raising enough money for the AI furnace.
Oracle will be.
https://www.cnn.com/2026/07/02/business/openai-trump-stake-intl
OpenAI giving the government stock for free would not be a bailout. Unless there's money moving in the other direction.
Maybe it's more like a bribe, but the question is what would OpenAI get for it.
Mostly this
Marketing and grift.
I think there’s a simpler explanation?
People—or at least the executives at many companies involved—believe these models are quite powerful and want to justify government controls so they have access to them before competitors do. Business as usual; capital building moats with regulatory capture!
The stated goal of having only a limited release of Mythos was to allow defenders to patch security bugs in important software before attackers get a chance to exploit them. I think that still applies?
I suppose you can twist that into “have access before competitors do” but it depends on what kind of competition you mean. We shouldn’t want to be neutral about the arms race between patching security bugs and exploiting them.
That stated goal sounds pretty flimsy from everything I have been reading.
Yeah, that would make a lot of sense too.
My guess is that the thieves (Anthropic and OpenAI) are afraid that they will be stolen from (via distillation). By releasing to a small set of true users they can build up their distillation defenses by getting a clear picture of what an actual usage distribution looks like vs a GA release that would definitely contain distillation traffic.