Notices by Kevin Beaumont (gossithedog@cyberplace.social)

Star Wars Zero Company is really good. It’s basically a great production budget version of X-Com.

A 16 year old boy hacked an internal Microsoft system called TITAN with 17 trillion records as it didn’t validate JWT signatures at all, accepted ‘admin’ as authentication without a password and was internet facing.

They paid him $5000 and told nobody about it.

https://blog.faav.net/how-i-couldve-accessed-17-trillion-microsoft-records

I’m starting to think everybody having financial and personal interests in portraying GenAI as an unstoppable monster isn’t a good idea.

ARTEX AI has been pulled off GitHub. I still have it. It’s a bag of shite.

It enumerates domains using subfinder, assets using nmap etc. It’s very basic, just a chain of open source projects.

It has hardcoded creds which allows anybody to use the webapp remotely.

It’s basically like being hacked by somebody off DALnet in 1998.

RE: https://infosec.exchange/@BleepingComputer/117417043223385492

There’s so much missing in these reports, which vendors are missing out.

Eg the main ‘attack’ was on a bank portal which had no form of MFA (at all). They also scraped the customer records by changing the account number in the URL.

@briankrebs of course he wrote a book on how to pay

@sawaba the devices had admin accounts with known passwords from prior breaches, and the attackers dumped the config including password hashes and then cracked the hashes

This thread is for those InfoSec leaders circulating the ‘OpenAI has solved maths’ posts. There’s a lot of them. Try asking experts in the field, rather than repeating corporate claims.

https://terrytao.wordpress.com/2026/09/11/a-severe-misalignment-of-ai-in-mathematics/

https://futurism.com/artificial-intelligence/mathematicians-openai-agents-proof-borderline-incomprehensible

https://techcrunch.com/2026/10/08/openais-math-solutions-arent-meeting-the-fields-standards-yet/

https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority_controversy

https://www.scientificamerican.com/article/did-openai-solve-the-wrong-navier-stokes-problem/

https://tech-insider.org/openai-math-papers-lean-verification-42-percent-2026/

https://www.newscientist.com/article/2592824-openai-mistranslated-mathematics-into-code-for-its-navier-stokes-proof/

https://www.theguardian.com/technology/2026/oct/07/openai-mathematical-findings-concerns

All credits to them for explaining this btw. The UK power station thing being targeted by Iran recently was actually foundational issues, same with the US water treatment incidents.

Because orgs don’t openly talk about what actually causes security incidents, it’s left a void of AI techbros filling executive ears.

RE: https://eupolicy.social/@bert_hubert/117412408358127056

@da_667 yep, it was during their ‘what’s next for Windows’ live stream two days ago.

They ended up with fewer viewers than before the presentation started, as they spent the entire hour talking about AI.. people roasted them in the comments and stopped watching.

The presentations had multiple typos in them, microphones didn’t work, the auto queue didn’t work properly, and the preorders didn’t work as the MS store website was already offline.

Preorder your Surface laptop even though we can’t spell our own product name