Nous Research Hits Back at GLM-5.3 Vulnerability Report, Calls Out Excessive Access Demands

NousResearch · x · 2026-08-15

Nous Research responds to reports that GLM-5.3 found 20 critical-to-high vulnerabilities in its Hermes agent, sarcastically suggesting they'd rather read message requests for hallucinated vulnerabilities, and arguing that Z.ai should share findings directly instead of demanding read access to entire GitHub orgs.

Related event: Nous Research Dismisses GLM-5.3's Report of 20 Critical Hermes Agent Flaws(2 posts)→

Original post →

More from Companies & People

Companies & People channel →