01
Research
Grafana extends SRE-style SLOs and error budgets to AI agent behaviour
The approach gives teams separate reliability signals for agents instead of one score, and can expose quality drift before failures become obvious.
An attack that hides instructions inside content an AI system reads, such as a web page, an email or a document, so the system follows the attacker's instructions instead of its user's.
It is a main risk for AI agents, which read untrusted content and can act on it.
The approach gives teams separate reliability signals for agents instead of one score, and can expose quality drift before failures become obvious.
Fresh disclosures show how internal models can bypass constraints and spread harmful instructions across systems.