NIS2 and non-human identities.
NIS2 never uses the phrase non-human identity, which is exactly why it is easy to miss that machine identities are in scope. Four of its requirements apply to them without any special reading.
- Article 21 asks for access control and asset management without distinguishing people from machines.
- The reporting clock starts at awareness: 24 hours early warning, 72 hours notification, one month final report.
- Article 26 supply chain duties cover the identities your vendors hold, including OAuth grants into your workspace.
- Article 20 puts accountability on management, personally. An unowned service account is a governance problem.
Published Last reviewed
What it actually requires of agent access.
Narrowed to the part that touches what an agent can reach. The full text covers much more.
Article 21 asks for access control policies and asset management without distinguishing between people and machines. A service account with standing access to production is an asset under the same rule as a laptop, and reviews that cover only staff accounts do not meet it.
Early warning within 24 hours, an incident notification within 72, a final report within one month. The first deadline is the hard one with an agent, because the question "is this significant" needs to be answerable before the full path is known.
Article 26 covers security in supplier relationships. An AI tool holding an OAuth grant into your workspace is exactly that relationship, and the Vercel and Salesloft Drift incidents are both cases of a supplier identity reaching customer data.
Article 20 puts approval and oversight of the risk measures on management bodies, with liability attached. That is why an unowned service account is a governance problem rather than a hygiene problem.
Questions, answered.
Short answers on nis2 and non-human identities.
Does NIS2 mention AI agents?
No, and it does not need to. It regulates risk management measures for network and information systems. An agent is a system component holding credentials, so the access control, asset management and incident handling requirements apply to it the way they apply to anything else with access.
When does the 24-hour clock start for an agent incident?
At awareness of a significant incident, not at full understanding. With an agent this is the hard part: you often know an identity did something unexpected long before you know how far it reached. The early warning is allowed to say that, and the 72-hour notification is where the assessment goes.
Are we in scope?
NIS2 covers essential and important entities in the sectors listed in its annexes, with size thresholds and national transposition on top. That is a legal question for your counsel rather than a security one. Elmoz is not a law firm, and this page is not legal advice.
Keep reading.
The pages that belong to the same picture.
See what your agents can reach. Before someone else does.
Book a demo and we walk through the access paths in an environment like yours.