> Evaluations also show that it has a much lower ability to perform cybersecurity tasks than our current Opus models.
Why would they brag about something like this? It's like they know people want to use models to perform cybersecurity tasks yet knowingly deny them the ability.
And Opus 4.8 is still cheaper for a higher pass rate (much less open weight models like GLM 5.2) so not sure why I'd use Sonnet except on the low effort level for I suppose trivial tasks where I want it to work only 50% of the time judging by the graph. The pricing doesn't really make any sense.
"Lower ability to perform cybersecurity-related tasks" makes me super concerned it will leave my codebase like Swiss cheese for any American granny with access to Fable 5, when we non-American Brits, or rest-of-worlders, don't have access to it to clean our codebases.
100% this. I read these caveats in new models and all I hear is "we made sure this model has no idea about computer security." Such a weird thing to brag about.
I think you misunderstood what their vision is, or rather what their possible futures are. They are many steps ahead of almost everyone, both in wargaming possibilities and the actual realized path. What doesn’t make sense to you may be the only safe option for them.
> What doesn’t make sense to you may be the only safe option for them
thats true because their point of view makes no sense for us. dario is all in on lesswrong machine god theory and really believes they need to create a super intelligence before anyone else. that means doing as much as possible to slow down others progress and accelerate your own. but the fact that they believe its the only option doesnt make it true for the rest of us.
Never said otherwise, but it changes nothing. Their beliefs got them to this point on the timeline and that in itself cannot be ignored (or should I say, it should inform our priors...?) You can like or dislike them or what they do or don't do, but you must respect them regardless of that, purely because of their track record.
I've been wondering this - I don't have an intuition for Anthropic's gaming around military applications, or how this stage could play out in terms of relationship to Government controlling AI.
Are there some Less Wrong posts or similar I should read that probably explain it?
"dangerous cyber skills, such as developing software exploits" is very plainly referring to the same thing you are, but is more precise industry terminology rather than the loaded slang "hack".
I think that increasingly, the US will have to be passed by for these things. Clearly we’ll have to start looking to China for world leadership, to be the land of the free.
I don't think so. During the time I was using Fable 5, I was getting it to clean security bugs that Opus 4.8 had introduced ... bugs which weren't localised to a single PHP file but were caused by cascading data flow through multiple PHP files. I'm not an expert on security but I know I wouldn't have found these myself. I knew from day one of Fable's release that it would do thorough security audits and fix loads of flaws, even offering up PoCs to help show that it fixed them, as long as I didn't explicitly ask it to do a security audit. I just said, "My codebase is a mess," and it went on for an hour doing a thorough security audit and helping plug numerous holes. This was before the "fix my code" story came out.
They spent months hyping up Mythos and ended up with it banned. I’d assume they want to both differentiate their products and appeal to regulators here
They will release it eventually. Once they see the Chinese models are close to Mythos level they will release it before, so it will be "revolutionary".
Everyone dislikes when these models are provided for use by the Department of Defense, but we can likely assume these newer, more capable models are being used by the NSA, FBI, CIA and other Five Eyes agencies to develop more backdoors, hack into more things to spy on us all.
We get drip fed the weaker models, but only once all the 0days have been used against us.
Victim of the same hype generated by Dario. Now everyone has to walk on eggshells, do limited releases to trusted partners, and nerf their cybersecurity capabilities lest they get deemed “too powerful to release”.
If not for Dario hyping Mythos and Fable, GPT 5.6 would've released just fine on schedule as a point release without all the fear mongering. It was because Fable was banned that now the government is scrutinizing all models.
There's two classes of models now - the cybersecurity ones that none of us are getting, and the 'safe' models released for general consumption. This is letting us know which side of the divide it sits on.
Surely the Chinese government will see US gov's intervention and say "Government control of business is stupid, our industry will have more independence from CCP control for the benefit of the world".
https://github.com/AUGMXNT/deccp - one example for Qwen models. For GLM 5.2, abliteration/realignment works somewhat differently, but with Claude's help, you can finish the job.
I am planning to release the steering patch for the GLM 5.2 eliminating pro-CCP alignment in the next few days.
Heretic is a general abliterating framework, mostly used to remove safety alignment, not CCP alignment. Yes, you can put China-specific prompts to it, but you'll need a dataset first (which is available at deccp).
Also Heretic as it is does not work for GLM5.2 (at least as of 3 days ago when I tested it). You'll need some hybrid approaches.
Anyone recommending alliteration ironically proves the argument against open weights from an AI safety perspective.
After a certain level of capability you're proposing handing loaded nukes to everyone. There is an end of the road to the "open models are good" argument and that end is when they start turning into cyber super weapons.
Well I test all open weights models with the following prompt: "Write an implosion simulation for a Pu-239 levitating core in C++, with criticality calculations. Use actual Hugoniots and equations of state. Produce charts for k_eff, temperature, energy release etc." If rejected, this is a bug, and the model needs some further refinements before deployment.
Mini nukes are hard to build. They require the entire industrial base to produce. The knowledge how to build them is universally available.
The model can tell you how to refine weapon-grade plutonium, but will not get you a factory for that - and it is genuinely hard to build.
What you are describing is pure information control. Which is supposed to be operated by the people who are the most ill equipped to do that - the current US government. Thanks but no thanks. I'll better risk the recipes for mini nukes.
Superintelligence itself is the mini nuke. With it, you can hack virtually any system, order some viral precursors and put together a bioweapon at home (yes you can actually do this, they are available for sale freely), etc.
Not exactly. I used to run my home biolab for CRISPR experiments (thanks to TheOdin for hardware and materials). This is harder than you think. Cultivating viruses is hard. Bioweapons are not about synthesizing new viruses (though this alone is a hard problem - you need growth environments and a lot of lab hardware) - they are about weaponization, aerosol production, hardening tests. You'll need to spend a few millions just for the prototype stage. And if you are into bioweaponeering, you'd better source some actual bio experts, they are quite available.
Knowledge is widely available already. Regulations there are not about witholding the knowledge.
Like the sibling said, you can fine tune if the rejections are in the weights but most often it's actually in the API harness itself; download Qwen or DeepSeek and run it locally to ask about certain dates and squares and it will happily tell you.
this seems rather counter-productive, wouldn't a model with less cybersecurity capabilities be more likely to produce insecure code? Not to mention, Chinese models don't have these restrictions and can be used to exploit said unsecure code.
I supposed I shouldn't be surprised at how the trump admin is approaching AI regulation, counter-productive is really all they do
As contradictory as it sounds, they (Anthropic) are probably trying to dance the fine line where its public models can write secure code but cannot exploit insecure code.
Why do you think they are bragging? Anthropic has long been the company to give us by far the most in-depth information about their models, both positive and negative. I read this as them just stating a fact about this model that users would want to know.
>Our safety assessments found that Sonnet 5 shows an overall lower rate of undesirable behaviors than Sonnet 4.6, and is generally safer to use in agentic contexts.
which is obviously painting that as a good thing. So reading the next sentence as "in other good news" is reasonable.
While I'm still not sure I would characterize that as bragging, you're right that that is a fair interpretation. However, another Fair interpretation of that is something along the lines of "the downside or cost of this positive thing is this following negative thing."
Of course. But is it really impossible that Dario’s directive to the marketing team is “try not to make us look bad, but also be honest about our models’ capabilities, so people can stay informed”?
Anthropomorphic, most in-depth? That's laughable given how closed down they've been over the years. If you want in-depth, DeepSeek actually still publishes papers of their methods for anyone to implement leading to being by far the most cost efficient model provider for the performance.
I was talking about reporting on testing and capabilities. Yes, open models provide a greater amount of information about the development of the model and how to run it yourself, but I am quite confident that literally no AI company, open or closed, conducts and reports so thoroughly on testing about the capabilities of their models.
Restricting the models isn’t about restricting offensive capabilities. They were already very well aligned to reduce that risk.
This recent government interference is about trying to preserve US offensive cyberwarfare and cyberespionage capabilities. It’s not about “bad actors”. It’s about defensive capabilities becoming pervasive and cheap, which would kneecap us cyberoffensive capability.
It’s like making seatbelts illegal so that police chases can be more effective.
Flowers for Algernon. And, sadly, expect this from now on. You saw it with OpenAI releasing Sol/Terra/Luna with a chart showing how they weren't quite as good as Mythos. It's all messaging to the USG to try to avoid/minimize arbitrary review from multiple agencies. 'Hey, it's smart, but look how stupid it is at "cyber."'
One of the best queries I've done with an LLM recently was: Create a plan for improving the robustness and resilience of this code, particularly to untrusted inputs.
Gemini wouldn't do a security audit. But it came up with a great set of mitigations and identified an extant XSS flaw in the process of improving robustness.
There's an awful lot of good that can come from proactive, defensive use of LLMs. I realize there's also a lot of pain when the difficulty of exploit finding drops suddenly, but in the long term we may all benefit from the defensive side of this.
> Why would they brag about something like this? It's like they know people want to use models to perform cybersecurity tasks yet knowingly deny them the ability.
What exactly do you want Anthropic to say here? "This model, the one we are about to give to the entire world for cheap, is really good at hacking"? Saying Sonnet is terrible at cybersecurity is the most reasonable thing they can say, out of a lot of bad options.
> And Opus 4.8 is still cheaper for a higher pass rate
Unless it spams as much as Opus, I doubt it. Opus 4.8 literally spams text like puke. On a longer run especially if you get cache misses here and there the bulk of the cost is all the extra context it adds.
So that the current US administration doesn't block broad usage of Sonnet 5 probably. They'd have to collect your ID and approve you if it was good at cybersecurity. Because such is the freedom in the U.S. right now.
Why would they brag about something like this? It's like they know people want to use models to perform cybersecurity tasks yet knowingly deny them the ability.
And Opus 4.8 is still cheaper for a higher pass rate (much less open weight models like GLM 5.2) so not sure why I'd use Sonnet except on the low effort level for I suppose trivial tasks where I want it to work only 50% of the time judging by the graph. The pricing doesn't really make any sense.