This post has been marked as a dupe, but it provides a lot more details than the other announcements of Fable's re-enablement provide:
> The export control directive on June 12 came after the government became aware of a report in which Amazon researchers had found a method of bypassing Fable 5’s safeguards: prompting it so that it identified a number of software vulnerabilities. In one case, the model produced code demonstrating how the relevant vulnerability could be exploited. Over the past two weeks, we have worked closely with the government and other partners, including Amazon, to review the report and evidence.
> Our testing confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities as Fable 5 did in the report. When it came to the demonstration of how to exploit the single vulnerability, every model we tested could produce the same demonstration as Fable 5 (including Claude Haiku 4.5, Sonnet 4.6, Opus 4.6, Opus 4.7, Opus 4.8, GPT-5.4, GPT-5.5, and Kimi K2.7).
This indicates three things:
1) WTF was Amazon thinking? Didn't their researches try the same thing in other models too before telling the CEO to tell the government it was dangerous (!?)
2) Anthropic - in particular Dario - really needs to learn government relations better. Most of the problems Anthropic has had with the government seem to stem from Dario's attitude rather than actual facts. (Eg, the DoD debacle seems to have ended up with OpenAI signing almost the same contract Anthropic already had, just worded differently)
3) The administration decision making is just wacky. In a normal administration they'd have actual policy documents you could look at to understand under what circumstances they think models have a problem. With this they just seem to make it up as they go, and the tools they use make no sense at all. If it is dangerous for cyber security reasons why would export controls make sense to use?
"For Pro, Max, Team, and select Enterprise plans, Fable 5 will be included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits."
Huge bummer. So what's the point of an anthropic subscription? Or is this just the end of subscriptions? Looking forward to gpt-5.6 (assuming we can still use our subscriptions there).
An Anthropic subscription is a 90%+ off discount for tech influencers to hype up an IPO (no different from companies directly paying Instagram influencers, except less overt).
Opus is at the same level as open weights models now. Okay, maybe a tiny bit better. So basically "nothing" - I don't see the point of using closed weights if there is an equivalent open weights model.
I really hope that one day we reach a point where we feel confident enough in the standards of care in upstream software, that we can get rid of these safeguards.
This isn't said out of naiveté or the idea that companies won't cheap out, but at some point – if access to models for defense is broadly available enough – we have to take a step back and say, "Aggressively insure your code against attack with AI on your side, because after <date> the other side will just _have_ AI."
I feel like something lost in a lot of the discussion around mythos and fable is that computer security absolutely has a substantial defender's advantage. It is indeed possible to ship e.g. surfaces that would be super resilient to attack (e.g. no unnecessary open surfaces, etc.) modulo category-shift attacks like RowHammer, etc.
Besides, just making sure that more people in the world actually have access to non-lobotomized models, this is _necessary_ if open source is (hopefully) going to continue to progress and if jailbreaks aren't totally vanquished.
I had already cancelled my subscription after finding the original Fable safeguards literally unusable (very basic chemistry, cryptography use cases), but with the false positives being admittedly worse now and the subscription not covering Fable for long I fail to see the point of consumer subscriptions now. Perhaps that's the goal, but it's a tough sell when Minimax M3 and GLM 5.2 are comfortably in Opus 4.5-4.8 territory but 1/5th-1/20th the price.
I've been wrestling for days with a recalcitrant Opus 4.8 unable to close the circle on an algorithm I'm sure, based on the three previous days' experience, Fable 5 will cut right through.
It’s nice to have access again, but I’m hesitant to include it in my workflow now. In fact I’d rather not use American models at all if I could.
I have the sense that this limited usage model will persist, Opus at its current token price will gradually be phased out, and eventually the cost of these models will be unsustainable and impractical.
I’d rather not play this game anymore and seek out more reliable providers which may provide less capable models but aren’t going to gradually kneecap my plan or rug-pull me.
Frankly I’m much less concerned with sheer model capability these days and more interested in harnesses. Claude Code is not the best on the market, and I think there’s a lot of exploring and learning to do in that arena.
The whole Fable debacle has dramatically changed my perspective on the market and what I want from it/who I should support.
Don't worry, now that Fable is freely available, a Chinese distill will be out a few months from now. To riff on Musk, the US AI industry is the bootloader for the Chinese AI industry.
Was there a use-case where you deployed Fable where you couldn’t use Opus in a non-interactive case? Perhaps Sonnet 5 will fill the gap because it’s better in agentic loops.
It’s so expensive, though. Useful interactively in coding agent but boy if you have a kind of task that Fable can do in an agentic loop that Opus can’t then you’re in a good place. Doubtless the frontier will move forward. These kinds of “problems of the future” are great to have solutions for.
And for an interactive agentic loop, not using Fable is just missing out on something. There’s no lock-in there.
I agree with this, the last three months have been complete nonsense. I'll take a less capable model with a workflow I know well over playing "what got silently removed this time" or "figure out the latest model's quirks".
The behavior of the american AI field/industry this year feels unstable and unsustainable, and I'm personally ready to wash my hands of this and find/build something reliable.
I mean that's the software playbook, get users addicted by offering it cheap and once you have them hooked, you raise the prices. There is no reason to think any non american company wouldn't follow this strategy. It's a different matter than the export control issue entirely and one that is systematic to software in general.
> There is no reason to think any non american company wouldn't follow this strategy.
This is generally true, but there are some providers offering inference at relatively stable prices. They aren’t Opus-tier models, but some appear to be at or close to Sonnet 4.5 or so. For much of the work I do, this is fine.
Essentially if you aren’t at the frontier, you can find cheaper tokens that aren’t about to be rug-pulled or decommissioned on a whim.
>Our testing confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities as Fable 5 did in the report.
This post has been marked as a dupe, but it provides a lot more details than the other announcements of Fable's re-enablement provide:
> The export control directive on June 12 came after the government became aware of a report in which Amazon researchers had found a method of bypassing Fable 5’s safeguards: prompting it so that it identified a number of software vulnerabilities. In one case, the model produced code demonstrating how the relevant vulnerability could be exploited. Over the past two weeks, we have worked closely with the government and other partners, including Amazon, to review the report and evidence.
> Our testing confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities as Fable 5 did in the report. When it came to the demonstration of how to exploit the single vulnerability, every model we tested could produce the same demonstration as Fable 5 (including Claude Haiku 4.5, Sonnet 4.6, Opus 4.6, Opus 4.7, Opus 4.8, GPT-5.4, GPT-5.5, and Kimi K2.7).
This indicates three things:
1) WTF was Amazon thinking? Didn't their researches try the same thing in other models too before telling the CEO to tell the government it was dangerous (!?)
2) Anthropic - in particular Dario - really needs to learn government relations better. Most of the problems Anthropic has had with the government seem to stem from Dario's attitude rather than actual facts. (Eg, the DoD debacle seems to have ended up with OpenAI signing almost the same contract Anthropic already had, just worded differently)
3) The administration decision making is just wacky. In a normal administration they'd have actual policy documents you could look at to understand under what circumstances they think models have a problem. With this they just seem to make it up as they go, and the tools they use make no sense at all. If it is dangerous for cyber security reasons why would export controls make sense to use?
"For Pro, Max, Team, and select Enterprise plans, Fable 5 will be included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits."
The party will be short-lived.
Huge bummer. So what's the point of an anthropic subscription? Or is this just the end of subscriptions? Looking forward to gpt-5.6 (assuming we can still use our subscriptions there).
An Anthropic subscription is a 90%+ off discount for tech influencers to hype up an IPO (no different from companies directly paying Instagram influencers, except less overt).
And yes, after the IPO.
Opus isn't nothing!
I think the 5x subscription is here to stay - I'd bet they make money on that from lots of people not using it.
The 20x is already unavailable in Teams plans.
Opus is at the same level as open weights models now. Okay, maybe a tiny bit better. So basically "nothing" - I don't see the point of using closed weights if there is an equivalent open weights model.
At least it is the end of endless arguments over subscription quotas.
You had two weeks to build up a queue of everything you wanted to do with it!
I really hope that one day we reach a point where we feel confident enough in the standards of care in upstream software, that we can get rid of these safeguards.
This isn't said out of naiveté or the idea that companies won't cheap out, but at some point – if access to models for defense is broadly available enough – we have to take a step back and say, "Aggressively insure your code against attack with AI on your side, because after <date> the other side will just _have_ AI."
I feel like something lost in a lot of the discussion around mythos and fable is that computer security absolutely has a substantial defender's advantage. It is indeed possible to ship e.g. surfaces that would be super resilient to attack (e.g. no unnecessary open surfaces, etc.) modulo category-shift attacks like RowHammer, etc.
Besides, just making sure that more people in the world actually have access to non-lobotomized models, this is _necessary_ if open source is (hopefully) going to continue to progress and if jailbreaks aren't totally vanquished.
I had already cancelled my subscription after finding the original Fable safeguards literally unusable (very basic chemistry, cryptography use cases), but with the false positives being admittedly worse now and the subscription not covering Fable for long I fail to see the point of consumer subscriptions now. Perhaps that's the goal, but it's a tough sell when Minimax M3 and GLM 5.2 are comfortably in Opus 4.5-4.8 territory but 1/5th-1/20th the price.
I've been wrestling for days with a recalcitrant Opus 4.8 unable to close the circle on an algorithm I'm sure, based on the three previous days' experience, Fable 5 will cut right through.
Yeah! I only knew it for 3 days but I was starting to fall in love!
Hooray! Glad everyone came to their senses and we can all get on with business.
I bet it'll continue to be messy at the frontier for the foreseeable future as society gradually wakes up to the consequences of strong AI.
I hope the redeployed Fable 5 is the same as the version they pulled without any substantially increased restrictions.
Per the article, the safety margin on the classifier is even worse than it was before. It sounds like the model itself hasn't changed.
It’s nice to have access again, but I’m hesitant to include it in my workflow now. In fact I’d rather not use American models at all if I could.
I have the sense that this limited usage model will persist, Opus at its current token price will gradually be phased out, and eventually the cost of these models will be unsustainable and impractical.
I’d rather not play this game anymore and seek out more reliable providers which may provide less capable models but aren’t going to gradually kneecap my plan or rug-pull me.
Frankly I’m much less concerned with sheer model capability these days and more interested in harnesses. Claude Code is not the best on the market, and I think there’s a lot of exploring and learning to do in that arena.
The whole Fable debacle has dramatically changed my perspective on the market and what I want from it/who I should support.
Don't worry, now that Fable is freely available, a Chinese distill will be out a few months from now. To riff on Musk, the US AI industry is the bootloader for the Chinese AI industry.
Was there a use-case where you deployed Fable where you couldn’t use Opus in a non-interactive case? Perhaps Sonnet 5 will fill the gap because it’s better in agentic loops.
It’s so expensive, though. Useful interactively in coding agent but boy if you have a kind of task that Fable can do in an agentic loop that Opus can’t then you’re in a good place. Doubtless the frontier will move forward. These kinds of “problems of the future” are great to have solutions for.
And for an interactive agentic loop, not using Fable is just missing out on something. There’s no lock-in there.
I agree with this, the last three months have been complete nonsense. I'll take a less capable model with a workflow I know well over playing "what got silently removed this time" or "figure out the latest model's quirks". The behavior of the american AI field/industry this year feels unstable and unsustainable, and I'm personally ready to wash my hands of this and find/build something reliable.
I mean that's the software playbook, get users addicted by offering it cheap and once you have them hooked, you raise the prices. There is no reason to think any non american company wouldn't follow this strategy. It's a different matter than the export control issue entirely and one that is systematic to software in general.
> There is no reason to think any non american company wouldn't follow this strategy.
This is generally true, but there are some providers offering inference at relatively stable prices. They aren’t Opus-tier models, but some appear to be at or close to Sonnet 4.5 or so. For much of the work I do, this is fine.
Essentially if you aren’t at the frontier, you can find cheaper tokens that aren’t about to be rug-pulled or decommissioned on a whim.
Nice! I go to the x20 again.
Discussion: https://news.ycombinator.com/item?id=48740771
"some routine tasks like coding and debugging will fall back to Opus 4.8."
disappointed but not surprised
>Our testing confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities as Fable 5 did in the report.
That is nice advertising for Kimi, huh?
In a way, but note that the government actors here have Opinions on the Chinese having capable models.