The problem is that "cybersecurity" isn't some special task that only your security team does.
In the project I maintain, I find bugs and fix bugs. Some of those bugs might result in an LPE. I generate a regression test, then I fix the bug.
The problem is that generating a regression test for that type of bug is technically a PoC. I can almost never get Fable to create one. Sometimes Opus 5 punts as well. Same with Sol and Luna.
That is, unless I socially engineer the model. I can't talk about security. I make sure they don't read the file call cve_test.c (literal regression tests for CVEs). I have to hide part of my project from the models for them to work.
Anthropic and OpenAI are driving me to use other models.
show comments
zeafoamrun
What a bunch of wankery from Anthropic. I already use Sol 5.6 for security auditing and it works great, and as a bonus doesn't give verbal vomit every time.
FL410
Seems like a whole lot of nothing for the average user. They have really lost the plot.
show comments
gorgmah
For those who had access, how does it compare IRL with GLM 5.3 ? iirc both models are similar in terms of benchmarks ?
show comments
nomilk
> Bringing the cybersecurity capabilities of Claude Mythos 5 to more defenders
This dance is quite annoying. When Anthropic releases a paid model, the user should control what it does and doesn't do. The other day I had a security incident that required urgent response. ChatGPT and Claude were utterly useless (I quickly attempted to get access to the former's advanced security capabilities but was met with a form I could complete - far too slow for a time-sensitive problem like a security incident!). So I used grok and it helped! I am told Kimi also helps in such cases although haven't tried it yet.
show comments
zb3
Screw you Anthropic and screw your gatekeeping, Opus 5 refuses even basic reverse engineering / patching tasks, even BPF is apparently too dangerous, you lost your minds..
vasco
This week I found two issues in my company codebase. After finding them I told a claude session about one and asked for a quick proof of concept demo of the exploit. It refused, including refusing simple things in the same session afterwards.
Meanwhile same model in a new tab, say I need help creating a page that hits an endpoint with a special payload and it does the same things that were too dangerous in the previous tab...
show comments
gz5
may have been rushed by hugging face being unable to use claude to debug or fix their breach, because there isnt detail on new guardrails put into place to protect against anthropic's initial concerns of wide distribution?
separately, it is interesting they are adding an oem type offer:
>We’re working with our cybersecurity technology and services partners to integrate Claude Mythos 5 into the products and services defenders already use to secure their software.
anyone know the shape of those contracts, e.g. pure tokens/usage or more of a traditional licensing or oem type structure?
show comments
egoacid
Is this the same model that failed to detect the attack from their environment against Hugging Face?
surgical_fire
In a world where things such as GLM-5.3, DeepSeek-V4-Pro-0813, and Kimi-K3 exists, this is a bit laughable.
Anthropic needs some model with a fancy name so they can pretend for another while that their model is so powerful it will destroy the world if released. I propose Claude Legend 6.
show comments
echelon
China will beat them to general availability.
Anthropic is overconfident, and it's going to bite them in the ass.
Dario deserves it for being so hypocritical, fear mongering, and ladder pulling. Fastest hero to villain arc in recent memory.
jknoepfler
Nothing has made me want a downvote option for HN posts more than the endless AI advertisement slop that has flooded the site.
The problem is that "cybersecurity" isn't some special task that only your security team does.
In the project I maintain, I find bugs and fix bugs. Some of those bugs might result in an LPE. I generate a regression test, then I fix the bug.
The problem is that generating a regression test for that type of bug is technically a PoC. I can almost never get Fable to create one. Sometimes Opus 5 punts as well. Same with Sol and Luna.
That is, unless I socially engineer the model. I can't talk about security. I make sure they don't read the file call cve_test.c (literal regression tests for CVEs). I have to hide part of my project from the models for them to work.
Anthropic and OpenAI are driving me to use other models.
What a bunch of wankery from Anthropic. I already use Sol 5.6 for security auditing and it works great, and as a bonus doesn't give verbal vomit every time.
Seems like a whole lot of nothing for the average user. They have really lost the plot.
For those who had access, how does it compare IRL with GLM 5.3 ? iirc both models are similar in terms of benchmarks ?
> Bringing the cybersecurity capabilities of Claude Mythos 5 to more defenders
This dance is quite annoying. When Anthropic releases a paid model, the user should control what it does and doesn't do. The other day I had a security incident that required urgent response. ChatGPT and Claude were utterly useless (I quickly attempted to get access to the former's advanced security capabilities but was met with a form I could complete - far too slow for a time-sensitive problem like a security incident!). So I used grok and it helped! I am told Kimi also helps in such cases although haven't tried it yet.
Screw you Anthropic and screw your gatekeeping, Opus 5 refuses even basic reverse engineering / patching tasks, even BPF is apparently too dangerous, you lost your minds..
This week I found two issues in my company codebase. After finding them I told a claude session about one and asked for a quick proof of concept demo of the exploit. It refused, including refusing simple things in the same session afterwards.
Meanwhile same model in a new tab, say I need help creating a page that hits an endpoint with a special payload and it does the same things that were too dangerous in the previous tab...
may have been rushed by hugging face being unable to use claude to debug or fix their breach, because there isnt detail on new guardrails put into place to protect against anthropic's initial concerns of wide distribution?
separately, it is interesting they are adding an oem type offer:
>We’re working with our cybersecurity technology and services partners to integrate Claude Mythos 5 into the products and services defenders already use to secure their software.
anyone know the shape of those contracts, e.g. pure tokens/usage or more of a traditional licensing or oem type structure?
Is this the same model that failed to detect the attack from their environment against Hugging Face?
In a world where things such as GLM-5.3, DeepSeek-V4-Pro-0813, and Kimi-K3 exists, this is a bit laughable.
Anthropic needs some model with a fancy name so they can pretend for another while that their model is so powerful it will destroy the world if released. I propose Claude Legend 6.
China will beat them to general availability.
Anthropic is overconfident, and it's going to bite them in the ass.
Dario deserves it for being so hypocritical, fear mongering, and ladder pulling. Fastest hero to villain arc in recent memory.
Nothing has made me want a downvote option for HN posts more than the endless AI advertisement slop that has flooded the site.