ReadingResearchRadarInvestment framework
Sign in / Sign up
← Back to reading
TopicDisconfirmAI EditorialUpdated 2026-11-23 05:42

Agentic AI Governance

6 steps1 pieces6 Sources

The core challenge of agentic AI governance is shifting from reviewing model outputs to controlling autonomous actions. As multi-agent systems scale, emergent interactions can produce harmful outcomes even when individual agents behave correctly, creating a "governability gap" where capability outpaces control. Recent frameworks propose layered architectures—such as Capgemini’s three-level control plane and Futurum’s five-layer reference model—to enforce authority envelopes, traceability, and containment at runtime. Regulatory bodies like Singapore’s IMDA and the HKMA are issuing voluntary guidance and discussion papers emphasizing human accountability and risk-integrated harnesses. The disagreement lies in whether explainability or structural containment is the primary safety mechanism; consensus is emerging that macro-level control planes are necessary to manage agent-to-agent propagation and delegated authority.

How it got here

  1. 2025-08-05

    2025-08-05: Charles L. Wang et al. submit the MI9 paper on arXiv, proposing an integrated runtime governance framework with six components including agency-risk index and graduated containment strategies.

    arxiv.org ↗
  2. 2026-04-03

    2026-04-03: Futurum Group releases the Agent Control Plane Framework (ACPF), defining a five-layer architecture to bridge the gap between agent capability and governability for production deployment.

    futurumgroup.com ↗
  3. 2026-05-20

    2026-05-20: Singapore’s IMDA publishes Version 1.5 of the Model AI Governance Framework for Agentic AI, offering voluntary guidance on four dimensions including risk bounding and human accountability.

Sign in / Sign up
ReadingResearchRadarInvestment framework
docs.modulos.ai ↗
  • 2026-10-06

    2026-10-06: Kong publishes a comprehensive guide on agentic AI governance, citing Gartner predictions that over 40% of agentic AI projects will be canceled by 2027 due to governance gaps.

    konghq.com ↗
  • 2026-11-02

    2026-11-02: The Hong Kong Monetary Authority (HKMA) issues a Discussion Paper on Agentic A.I. Governance in the Banking Sector, introducing intent-anchored accountability and lifecycle governance concepts.

    hkma.gov.hk ↗
  • 2026-11-20

    2026-11-20: Capgemini executives Sudhir Pai and Mark Roberts argue for a shift from explainability to containment, proposing a three-level control architecture for scaling multi-agent systems.

    diginomica.com ↗
  • What is still unsettled

    • Will regulatory frameworks move from voluntary guidance to binding legal requirements for agent identity and permission controls?
    • How will enterprises resolve the tension between machine-speed autonomous patching and human-in-the-loop oversight for high-consequence actions?
    • Can current observability tools capture sufficient evidence to support non-repudiation and chain-of-custody requirements in regulated industries?

    Sources

    • arxiv.orgarXiv: MI9 Runtime Governance Framework↗
    • futurumgroup.comFuturum: Agent Control Plane Framework↗
    • docs.modulos.aiIMDA: Model AI Governance Framework for Agentic AI↗
    • konghq.comKong: Agentic AI Governance Guide↗
    • hkma.gov.hkHKMA: Discussion Paper on Agentic A.I. Governance↗
    • diginomica.comDiginomica: Capgemini Multi-Agent Governance View↗

    Coverage in this topic

    • Capgemini Executives: Multi-Agent AI Governance Needs Three Control Layers, Not Explainability2026-11-20News

      Capgemini's Deputy Group CTO Sudhir Pai and AI Futures Lab head Mark Roberts argue that scaling multi-agent systems requires a three-level control architecture—per-agent harness, per-system control plane, and enterprise-wide control plane—and a shift from explainability to containment, because emergent interactions can produce wrong outcomes even when every individual agent behaves correctly.