The International Telecommunication Union has opened a new front in the effort to govern artificial intelligence, and building automation professionals should be paying close attention. On July 9, 2026, ITU announced the formation of a Focus Group on Trust and Identity for Humans and Agentic AI, a body tasked with developing the global standards needed for autonomous AI agents to interact safely with people and with one another. The group was launched at the AI for Good Global Summit and will report to ITU-T Study Group 17, the organization’s home for security standards.
The timing is not incidental. AI has moved past chatbots and copilots into systems that plan, decide, and act with limited human involvement. ITU’s own framing captures the shift plainly: agentic AI promises real productivity gains, but it also opens the door to agents impersonating people or organizations, or taking unauthorized action across the interconnected systems they touch. The Focus Group’s job is to build the identity and trust scaffolding that keeps autonomy accountable, particularly in domains ITU specifically calls out, such as financial transactions and critical infrastructure.
Buildings are critical infrastructure. That single fact is why this initiative belongs on the desk of every controls engineer, integrator, and building owner watching AI move into their systems.
Agentic AI is already arriving in the field
This is not a hypothetical future problem. Honeywell introduced Forge AI capabilities that integrate with Niagara-based building management workflows. Johnson Controls continues expanding generative AI capabilities within OpenBlue, and Trane has incorporated autonomous optimization technologies into its platform portfolio. Together, these developments suggest that major BAS vendors are actively investing in increasingly autonomous software capabilities. Independent research from Memoori tracks all of this as a genuine competitive race among the major building automation incumbents, not a marketing exercise.
At Niagara Summit 2026, sponsors and partners described agent-based AI as something already turning building data into “autonomous, intelligent operations,” with edge and cloud infrastructure built specifically to support it. Many vendors are introducing increasingly autonomous software capabilities into commercial building platforms before international standards for AI identity and trust have matured.
That gap is exactly what ITU’s Focus Group intends to close, and the building automation sector has more relevant experience than most industries walking into this conversation.
BACnet/SC already solved a piece of this puzzle
Building automation does not have to invent the concept of device identity and trust from scratch. BACnet Secure Connect, the security-focused addendum to the BACnet standard, requires participating BACnet/SC devices to authenticate using X.509 digital certificates before establishing secure communications. Two devices authenticate each other by validating those certificates, and a device without one is simply refused a connection. It is, in effect, an identity and trust framework for machines that has been live in commercial buildings for several years, well before “agentic AI” entered the vocabulary.
ITU’s proposed reference architectures for identity, trust lifecycle models, and interoperable credentials are attacking the same underlying problem BACnet/SC addresses at the device layer, except now applied to autonomous software agents that reason and act rather than passive controllers that report and respond. The BAS industry’s certificate-based trust model, and the broader work already underway on finer-grained authorization and zero-trust approaches within the BACnet standard, is a working precedent ITU’s group would do well to study, and one building automation vendors and standards bodies are well positioned to bring to the table.
What this means in practice for building automation
A few implications are worth tracking as this Focus Group takes shape:
Agent identity becomes a building system requirement, not just an IT concern. As HVAC optimization agents, fault-detection agents, and energy-management agents begin negotiating and executing actions across BAS platforms, cloud services, and utility interfaces, the building industry will need clear answers to the same questions ITU is posing at the global level: which agent is acting, on whose authority, and within what boundaries.
Interoperability standards will matter as much as security standards. ITU’s mandate includes common terminology, reference architectures for agent discovery, and interoperability mechanisms for credentials. Building automation already wrestles with multi-vendor interoperability through BACnet, Niagara, and semantic tagging efforts like Project Haystack. Agentic AI adds a new interoperability layer on top: agents from different vendors that need to discover, trust, and coordinate with one another within a single building or portfolio.
Human oversight is being designed in, not bolted on. ITU’s announcement explicitly emphasizes maintaining meaningful human control over autonomous AI systems, particularly in financial transactions or critical infrastructure. That principle aligns closely with long-standing building automation practices that retain operator oversight for safety-critical functions. That framing aligns closely with how the building automation industry already approaches safety-critical overrides in life-safety and critical infrastructure systems.
Governance frameworks increase confidence to deploy; they do not just add compliance overhead. That is a lesson building automation learned with BACnet/SC itself. Facility owners and integrators did not treat certificate-based authentication as red tape; it became the foundation that let them extend BAS networks onto the public internet and into the cloud with confidence. The same pattern is likely to hold for agentic AI governance in buildings.
A seat at the table
ITU’s Focus Group holds its first meeting in Paris in November 2026, followed by a second session in Geneva in January 2027, and it is explicitly open to technical experts as well as specialists in policy, law, and regulation. Given how much of the underlying identity and trust groundwork already exists in BACnet/SC, and how quickly agentic AI is being layered onto BAS platforms in the field, building automation’s standards bodies, vendors, and integrators have a genuine stake in this process and real expertise to contribute.
The building automation industry has spent the better part of a decade learning how to establish trust between machines on a network. That experience did not disappear when the machines in question began to reason and act on their own. The BAS industry’s experience with certificate-based authentication offers a practical example of how identity and trust can be implemented in operational technology environments. It is experience that could inform broader discussions around AI identity in critical infrastructure.
Sources and Further Reading:
“ITU launches global standards initiative to build trust in AI behaviour and identity,” July 9, 2026: https://www.itu.int/en/mediacentre/Pages/PR-2026-07-09-focus-group-agentic-AI.aspx
ITU Focus Group on Trust and Identity for Humans and Agentic AI: https://www.itu.int/en/ITU-T/focusgroups/tida/Pages/default.aspx
Cimetrics, “What is BACnet Secure Connect and Why It Matters”: https://cimetrics.com/what-is-bacnet-secure-connect-and-why-it-matters/
ASHRAE, “BACnet Secure Connect: A Secure Infrastructure for Building Automation” (whitepaper): https://www.ashrae.org/file%20library/technical%20resources/bookstore/bacnet-sc-whitepaper-v15_final_20190521.pdf
Memoori, “Honeywell AI Strategy in Smart Buildings: 2026 Assessment”: https://memoori.com/honeywell-ai-strategy-smart-buildings/
Memoori, “AI Strategies of Smart Building Incumbents: 2026 Compared”: https://memoori.com/ai-strategies-smart-building-incumbents/
Related Reading from AutomatedBuildings.com
The Five-Dollar Widget: What Happened to Trust on the Way to the Digital Age
https://www.automatedbuildings.com/2026/07/the-five-dollar-widget-what-happened-to-trust-on-the-way-to-the-digital-age/
When Buildings Govern Themselves
https://www.automatedbuildings.com/2026/03/when-buildings-govern-themselve/
How Digital Certificates are Used in BACnet/SC
https://www.automatedbuildings.com/2026/02/how-digital-certificates-are-used-in-bacnet-sc/
The BACnet/MSI Semantic Gap: Why Independent Integrators Hold the Key to Unlocking Agentic Buildings
https://www.automatedbuildings.com/2026/06/the-bacnet-msi-semantic-gap-why-independent-integrators-hold-the-key-to-unlocking-agentic-buildings/
Why BAS Engineers Need a Cybersecurity Design Guide
https://www.automatedbuildings.com/2026/05/why-bas-engineers-need-a-cybersecurity-design-guide/
