Track 2

Agent Governance and Observability: Mitigating Risk and Eliminating Rogue AI Cost Centres

Scaling use cases means scaling agents: IDC research shows that organisations expect to double the number of distinct agent types in 2026 alone. The more agents you run in production, the wider the gap between what your governance frameworks were built for and what they are now being asked to cover. Unmonitored agents create compliance exposure and unaccountable costs, while unsanctioned workloads silently drain budgets. “The immediate threat from agents is self-inflicted harm due to poorly governed agentic deployments,” says recent IDC research.

This track addresses the governance infrastructure required to scale safely, covering identity and access controls, observability frameworks, and audit capabilities that keep agents within defined boundaries and leadership fully informed.

But governance is only half the equation. The same forces driving agentic adoption inside your organisation are arming the adversary outside it. Attackers are now using AI to probe, exploit, and move at machine speed — automating reconnaissance and scaling social engineering beyond what human teams could mount. “Threat actors, increasingly augmented by their own AI systems, are already compressing attack life cycles from days to hours and, soon, from hours to minutes,” says IDC.

Defending an expanding agent estate demands a posture that is equally automated, continuous, and intelligence-led. Organisations need to scale with confidence by treating governance and cyber resilience as two sides of the same strategic imperative.