Every security program accepts risk. A legacy system cannot be patched until the replacement ships. A vendor without single sign-on is too embedded to drop this year. A finding from the audit will be fixed next quarter, not this one. None of that is a failure. Accepting risk deliberately is what risk management is.
The failure is accepting it without a record. After an incident, investigators, auditors, insurers, and sometimes regulators reconstruct what was known and who decided. If the only trace is a thread in a chat tool and a memory of a meeting, a deliberate business decision reads exactly like a gap nobody noticed, and the person left holding it is usually whoever raised it.
This brief covers the record itself. What it means legally in your situation is a question for counsel.
What the public references say
Risk acceptance is a standard response, not an admission. NIST IR 8286 Rev. 1, published in December 2025, lists four responses to a negative cybersecurity risk: accept, transfer, mitigate, and avoid. It describes acceptance as appropriate for risks within tolerance, needing no further action except monitoring, and it is explicit that risks marked as accepted still need to be measured to ensure they stay within tolerance.
NIST CSF 2.0 makes the record an outcome: changes and exceptions are managed, assessed for risk impact, recorded, and tracked (ID.RA-07). And the NIST Risk Management Framework, SP 800-37 Rev. 2, builds expiry into acceptance: an authorization decision carries terms and conditions and a termination date set by the authorizing official.
Three references, one shape. Acceptance is legitimate, it is recorded, it has an owner with authority, and it ends.
The nine fields
| Field | What to write |
|---|---|
| Risk | One or two sentences a non-specialist understands. What could happen, to what, and how. |
| Scope | The systems, data, or business processes affected. |
| Exposure against tolerance | Current likelihood and impact, and whether that is inside or outside the agreed tolerance. |
| Reason | Why this is not being fixed now: cost, dependency, timing, business need. |
| Compensating controls | What reduces exposure in the meantime, and how you know they work. |
| Accepting owner | A named individual, not a team. |
| Authority | Why this person can accept this level of risk: role, delegation, or committee approval. |
| Expiry | A date. Not "until fixed". |
| Reopen triggers | Events that reopen the acceptance early: an incident, a scope change, a new exploit, a control failure. |
These map closely onto the register elements NIST IR 8286 Rev. 1 describes, which include risk description, likelihood, impact, exposure rating, response type, response description, risk owner, and status. Using that vocabulary makes the record legible to auditors who already know it.
Who signs
The business owner of the affected system or process signs. Not the security team.
That is not about avoiding responsibility. The owner holds the benefit of not fixing the risk now, whether that is budget, a delivery date, or a customer commitment, so the owner holds the decision. Security's role is to describe the risk accurately, recommend, and record dissent if it disagrees. A record where security accepted a risk on behalf of the business is the weakest version, because it shows the person with the least authority making the decision.
Set acceptance authority by exposure. A risk inside tolerance can be accepted by the system owner. A risk outside tolerance needs a more senior signature, and a risk well outside tolerance needs the executive or committee that set the appetite. Write those levels down once, in policy, so the question of who can sign is never argued case by case.
Why the obvious answer is wrong
The obvious answer is a ticket in the risk register with a status of "accepted". That records the state, not the decision. It usually lacks the reason, the authority, and the expiry, which are the three fields that matter most later.
The second obvious answer is to avoid writing acceptances down at all, on the theory that a written record of a known risk is a liability. The practical problem is that an unwritten known risk is still known; it is simply known without evidence that anyone decided anything about it. Whether a written record helps or hurts in a particular dispute is a question for counsel. As a matter of practice, programs under audit are expected to show the record.
The discriminating variables
Is it inside or outside tolerance?
The test: compare the exposure with the board-approved tolerance. Inside, the system owner can usually accept. Outside, the signature moves up. If no tolerance exists, every acceptance is a judgement call with no anchor, and that is worth fixing first.
Is the expiry real?
The test: is there a date, and does something happen on it? An expiry that nobody reviews is decoration. Put the dates in the same calendar as the quarterly board report, and report expiring acceptances there.
Do the compensating controls exist?
The test: can you show evidence that each compensating control is operating, today? A compensating control that was planned but never deployed turns the record against you.
Has it been renewed more than once?
The test: count renewals. A risk accepted three times in a row is no longer a temporary acceptance; it is a permanent one, and it should go to the authority that can accept permanent risk, with the full history attached.
Decision table
| Situation | Who accepts | Expiry |
|---|---|---|
| Inside tolerance, compensating controls in place | System owner | Up to a year |
| Outside tolerance, fix scheduled | Senior executive for the area | The fix date, plus a short margin |
| Well outside tolerance | Executive or committee that set the appetite | Short, reviewed every quarter |
| Renewed more than twice | Escalate one level | Reset, with history attached |
| Security disagrees with the acceptance | Owner accepts; dissent recorded | As above |
Recording dissent
If security recommends against accepting, the record should say so, in a sentence, with the date. That is not a tactic for later. It is an accurate account of the decision, and it lets the owner see clearly what they are signing. Most owners, faced with a written dissent, either ask for more information or choose a shorter expiry, which is the record doing its job.
What changes the answer
A regulatory or contractual control. Some controls are obligations, not risk choices. A business owner may not be able to accept a gap in something a regulator or a customer contract requires. Treat those as escalations to counsel and the obligation's owner, not as ordinary acceptances.
An active incident. An acceptance made during an incident is a response decision. Record it in the incident record, with the same fields, and revisit it in the post-incident review.
Public statements. If a trust page, questionnaire answer, or filing describes a control that an open acceptance says is missing, the two records contradict each other. Find those contradictions before someone else does, and take them to whoever owns the public statement.
Next steps
- Pull every open acceptance in your register and check each for the nine fields. Most will lack authority and expiry.
- Write acceptance authority levels into policy, tied to the board's tolerance.
- Add expiring acceptances to the quarterly board report.
- Keep a decision log for the decisions that are not risk acceptances but will still be reconstructed later.
- Start from the risk acceptance record template.