AINews 'Black October' roundup clusters AI-linked security disclosures
Original title: [AINews] Black October
The digest says many incidents were found earlier and disclosed together, while exploitation scale is disputed.
Latent Space's AINews recap calls a roughly 48-hour cluster of disclosures and breaches "Black October" and treats it as historically significant for AI security. It lists claimed issues in PHP, GitLab, Node.js, LiteLLM, Railway, Zeabur, eMudhra and Quest Diagnostics, plus Microsoft's attribution of a ConnectWise compromise to Storm-0940. It also covers a METR and Redwood on-site review of OpenAI agent monitoring, including reported bypasses and redactions. Several claims are disputed in the same roundup.
Unconfirmed: Figures and attribution in the roundup are secondhand, and PHP discovery credit, bank-agent claims, and large-scale exploitation are disputed.
- GitLab's flaw is described as CVSS 10, patched in 19.5.1, affecting releases since 2014.
- Zeabur is said to be shutting down after 321,810 accounts were breached.
- Microsoft attributes the ConnectWise compromise to Storm-0940 using ARTEX against 2,250-plus tenants.
- METR and Redwood spent three weeks on-site; the report was cut from 61 to 38 pages.
- One commenter says only about 4 of roughly 16 major incidents occurred this week.
- Threat actors
- Storm-0940
- Malware
- ARTEX
- Organizations
- PHP FoundationMETRRedwood ResearchCrowdStrike
- Countries
- ChinaArgentinaUnited States
- AI models
- ClaudeClaude Mythos
From the original article
14 min read at latent.space
If you are reading this, you’re receiving this from our new provider, Beehiiv! We are making progress on our redesign — if you have a friend that didn’t receive AINews, please find us in Spam filter and move us out to your Inbox… thank you!
By swyx .ai · latent.space
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.