x.comthreadAGTP (@AGTPinsights) on X
Thread on a 3 November 2026 exchange in which Primcloud's Josh Manders accused Railway employee Jake Runzer of trying to breach his infrastructure. Runzer replied that Railway's agentic testing looked "off the rails", that models are "getting a bit too good", and that Railway will ban non-manual external testing. It follows Railway's own breach of 20 to 21 October 2026. No logs or formal statement have been published.
TL;DR Reports Railway's apology after its AI testing agents allegedly probed a rival cloud, with no logs published.
Organisations named here
- Jake Runzer
- Josh Manders
- Primcloud
- Railway
Technologies named here
- Railway
- Railway internal systems
- Primcloud
- Railway agentic testing platform
Techniques named here
- Security probing by autonomous AI agents
Concepts named here
- Agent owner accountability
- Log-based verification
- Unauthorized external security testing
Related in AI Security
blog.talosintelligence.comtoolIntroducing CAIRN: Frontier tracking for AI-integrated malware (Cisco Talos)
github.comadvisoryFlowise MCP Security Bypass Enables RCE
github.comrepoYesWeHack Claude Kit
jfrog.comarticle