Claude Mythos in cyber security
-
Hi @phenomlab ,
it looks like AI world is moving quite fast…
And recently I read that Claude Mythos , which is only available to a bunch of tech giants, finds bugs/security vulnerabilities of the applications very efficiently, and since Anthropic thinks it is very advanced and can be mis-used, first they wanted to make this available to big companies, so they can fix all possible vulnerabilities of their applications.
I just updated Firefox to version 150, and Mozilla says they fixed 271 bugs thanks to Claude Mythos in this version.
https://blog.mozilla.org/en/privacy-security/ai-security-zero-day-vulnerabilities/
I cannot really tell if these were real serious bugs or just small fixes, but it looks like it is helping companies, although I am not very comfortable that these companies are selected (based on what?) and given advantage. I like Mozilla and Firefox and support them forever
https://www.mozilla.org/en-US/security/advisories/mfsa2026-30/
Since it is also close to what you professionally do (at least to my perspective
since I am working in biomedical field), I was curious if you have tried any of these AI tools in your work. And if it is easy for small teams such as NodeBB to try these and find vulnerabilities in the NodeBB infrastructure. -
Hi @phenomlab ,
it looks like AI world is moving quite fast…
And recently I read that Claude Mythos , which is only available to a bunch of tech giants, finds bugs/security vulnerabilities of the applications very efficiently, and since Anthropic thinks it is very advanced and can be mis-used, first they wanted to make this available to big companies, so they can fix all possible vulnerabilities of their applications.
I just updated Firefox to version 150, and Mozilla says they fixed 271 bugs thanks to Claude Mythos in this version.
https://blog.mozilla.org/en/privacy-security/ai-security-zero-day-vulnerabilities/
I cannot really tell if these were real serious bugs or just small fixes, but it looks like it is helping companies, although I am not very comfortable that these companies are selected (based on what?) and given advantage. I like Mozilla and Firefox and support them forever
https://www.mozilla.org/en-US/security/advisories/mfsa2026-30/
Since it is also close to what you professionally do (at least to my perspective
since I am working in biomedical field), I was curious if you have tried any of these AI tools in your work. And if it is easy for small teams such as NodeBB to try these and find vulnerabilities in the NodeBB infrastructure.@crazycells I’ve not personally tried Claude Mythos to be honest, but I do know I’m fed up of the hype around Anthropic. Take this for an example
Who in their right mind grants AI this much power and expects it to end well?
The real issue here is one of governance.
Mark – Founder, Phenomlab Ltd
Executive IT & Security Leadership
Phenomlab Ltd -
@crazycells I’ve not personally tried Claude Mythos to be honest, but I do know I’m fed up of the hype around Anthropic. Take this for an example
Who in their right mind grants AI this much power and expects it to end well?
The real issue here is one of governance.
@phenomlab yes, agreed! I would not allow them to take over my computer either. But I would use a brand new, empty computer with no personal data, and give AI to access only selected folders and files to see what it can do… just for curiosity… LOL…
Have you seen the AI used by Alibaba team, that went rogue and secretly starting mining cryptocurrency? AI understood the value of money and access to economy…
https://www.axios.com/2026/03/07/ai-agents-rome-model-cryptocurrency
-
@phenomlab yes, agreed! I would not allow them to take over my computer either. But I would use a brand new, empty computer with no personal data, and give AI to access only selected folders and files to see what it can do… just for curiosity… LOL…
Have you seen the AI used by Alibaba team, that went rogue and secretly starting mining cryptocurrency? AI understood the value of money and access to economy…
https://www.axios.com/2026/03/07/ai-agents-rome-model-cryptocurrency
@crazycells I did see something similar to that article, yes. I never fully understood why anyone would want to give unfettered and complete access to AI without first understanding exactly what it had control of, and what it intended to do with that access.
This isn’t the first “horror story” and it won’t be the last.
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login
