AI is rapidly shifting from a passive tool that answers questions to an active agent that completes complex tasks on its own. As these systems gain more autonomy, traditional ways of managing risk and compliance are falling short.
Simply instructing an AI to "respect user privacy" or "follow company policy" in plain English is no longer enough. Everyday language leaves too much room for interpretation, leading to unpredictable results and making it nearly impossible to audit the AI's decisions.
To safely scale these autonomous systems—whether in business platforms or scientific research—the rules of the road must be built directly into the technology. This means shifting to "policy-as-code." Instead of relying on natural-language suggestions, organisations must translate data rights, legal duties, and safety constraints into strict, machine-readable rules. By embedding these rules directly into the data and the tools the AI uses, the system is forced to verify its boundaries before taking any action.
Join our panel of researchers, system architects, and governance experts to discuss how we can turn high-level policies into hard-coded reality.
What we’ll cover:
- The enforcement gap: Why giving AI plain-language instructions is an unreliable safety guardrail, and what stricter, code-based rule enforcement actually looks like.
- Real-world research: Early insights from the Open Data Institute’s live research on governing autonomous AI in enterprise environments.
- The legal translation: How corporate legal teams, compliance officers, and data councils are turning complex contracts and privacy laws into actionable software code.
Chaired By Dr June Brawner, Head of Research, the ODI