Shifting Recon Priorities: AI/LLM Infrastructure.
My passive sensors are capturing targeted probing on OpenAI/Anthropic compatible endpoints. This isn't generic scanning; it's a search for exposed LLM proxies.
Technical Details:
Agent: FastScan/1.0
Target: /v1/chat/completions, /v1/models
Observation: Attackers are moving from traditional web exploits to hunting for "Model-as-a-Service" misconfigurations to monetize compute or leak system prompts.
Is anyone else seeing FastScan activity in their network background noise?