AI Governance in Quality Engineering: From Policy to Practice
As AI becomes embedded across the SDLC, organizations need to move beyond high-level policies and make governance operational. This session explores what practical AI governance looks like from a Quality Engineering perspective. It covers how teams can set clear boundaries for AI systems, define accountability and risk acceptance, continuously test and monitor AI behavior, and use metrics and evidence to determine whether AI is delivering value safely.
Using recent real-world AI incidents and a practical agentic QE example built around Playwright, MCP, Claude, and self-healing automation, the session shows how governance can be built directly into engineering workflows without becoming a final approval gate that slows innovation.
The session is centered around a simple three-part framework: Set the Boundaries, Check the Behavior, Measure the Value.
Three key takeaways
1. Move governance from policy to practice
Effective AI governance is not just policies and compliance reviews. It requires clear guardrails, ownership, risk acceptance, monitoring, testing, and evidence built into how AI systems actually operate.
2. Govern AI based on behavior and risk
As AI becomes more agentic, teams need to understand what systems can access, change, and act on, then continuously validate that they remain within those boundaries. Human accountability remains essential, while the level of human intervention should be proportional to risk.
3. Quality Engineering can become the assurance layer for AI
QE can translate governance expectations into testable controls, evaluations, monitoring, and measurable evidence, helping organizations determine not only whether AI works, but whether it can be trusted to take on greater autonomy.
