Recovery fails because the same authority can corrupt the evidence, delete backups, or rewrite the restore point. In that situation, a backup no longer proves trustworthiness. Teams need a recovery path that sits outside the compromised admin domain and can be restored without relying on the attacker-controlled control plane.
Why Recovery Breaks When One Authority Controls Both Sides
Recovery depends on separation of trust. If the same production identity can also reach the recovery copy, the attacker can alter logs, rotate keys, delete snapshots, or replace the restore point before defenders notice. The result is not just loss of data, but loss of assurance: the backup may still exist, yet it can no longer be trusted as a clean source of truth. That is why recovery design has to assume the control plane can be hostile.
Practical recovery design therefore treats backup systems as a distinct trust boundary, with independent credentials, independent administrative review, and restoration paths that do not inherit production authority. That separation matters most when identity is already compromised, because restore is then a security decision, not only an operations task. The Ultimate Guide to NHIs shows how often credentials stay valid and over-privileged long enough to make that boundary matter in real incidents. In practice, many teams discover backup corruption only after they need the restore point most.
How Recovery Architecture Holds Up in Practice
A resilient recovery path is built so that production compromise does not automatically transfer into backup compromise. That usually means separate identities, separate administrative consoles, and separate storage or snapshot management, with a limited set of operators able to approve destructive changes. It also means the restore path must be testable without requiring the same credentials that protect daily operations.
- Keep backup administration outside the production admin domain.
- Use immutable or tamper-evident copies where the platform supports it.
- Protect backup credentials with tighter scope than production privileges.
- Record restore-point changes in a system the production identity cannot rewrite.
- Test that recovery still works after revocation, rotation, and account lockout.
This is not only about preventing deletion. An attacker who can control both sides may also poison the recovery set by choosing a compromised point in time, hiding the original intrusion, or staging a delayed re-compromise after restore. That is why restore procedures should verify integrity, timestamps, and known-good configuration before the system returns to service. The compromise path is especially dangerous when backup tooling, cloud control planes, or shared secrets let one administrative role reach everything. The Anthropic report on the first AI-orchestrated cyber espionage campaign is a useful reminder that attackers increasingly automate actions once they obtain high-value access, which makes rapid control separation more important. These controls tend to break down when backup operations are merely a permission inside the same tenant, account, or admin plane as production.
Common Variations and Edge Cases
Tighter isolation improves trust, but it also increases operational overhead, so teams have to balance recoverability against administrative convenience.
Some environments rely on snapshots, some on offline copies, and some on object-lock or immutable storage. The right answer depends on how quickly the business must recover, how much data loss is acceptable, and whether the platform truly prevents a compromised production identity from rewriting the recovery set. In regulated or high-availability environments, the standard is usually not “a backup exists,” but “a backup can be restored independently under hostile assumptions.”
Hybrid environments create another edge case: the production identity may be compromised in one cloud or on-prem domain while the recovery copy sits elsewhere, yet federation, shared secrets, or delegated administration can silently rejoin the trust boundary. The strongest designs assume the attacker will look for that hidden bridge first. The MITRE ATT&CK Enterprise Matrix is useful here because it helps teams think in terms of privilege abuse, defence evasion, and post-compromise actions rather than only initial access. The decision point is simple: if the same authority can rewrite the evidence and the recovery point, the restore path is already unsafe.
Risk and Threat Considerations
The material risk is loss of recoverable trust, not just loss of availability. When production identity and the recovery copy share authority, an attacker can turn backup infrastructure into a persistence layer, a deletion target, or a false source of truth.
Failure mechanism: The attacker abuses legitimate administrative access to delete snapshots, alter retention, rotate keys, poison restore points, or conceal compromise history, then uses the same access path to block or subvert recovery.
Impact: Defenders may restore compromised data, lose evidence needed for incident response, or find that no trustworthy clean recovery point remains, which can extend downtime and increase business and compliance exposure.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
MITRE ATT&CK address the attack and risk surface, while NIST CSF 2.0 and CIS Controls v8 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST CSF 2.0 | RC.RP — Recovery Planning | Recovery depends on a separate, testable restore path after compromise. |
| PR.AC — Identity Management, Authentication and Access Control | Shared authority over production and recovery is the core trust failure. | |
| Recommendation — Design and test recovery so restore can proceed after production compromise. Separate and restrict access so production identity cannot control recovery copies. | ||
| CIS Controls v8 | 8.10 — Data Recovery | Recovery copies must remain usable and trustworthy under hostile conditions. |
| Recommendation — Maintain and verify restore points that survive account compromise and tampering. | ||
| MITRE ATT&CK | T1485 — Data Destruction | Attackers may delete or corrupt backups once they gain administrative control. |
| T1070 — Indicator Removal on Host | The same access path can be used to hide evidence before recovery. | |
| Recommendation — Hunt for destructive actions against snapshots, backups, and restore points. Preserve immutable logs and evidence outside the compromised admin domain. | ||
Practitioner Guidance
What to prioritise: Treat backup administration as a separate recovery control, not a sub-permission of production operations. If production admins can both destroy and restore the same asset set, your blast radius is larger than the control description suggests.
What to verify: Confirm that a restore can be executed after production credentials are revoked, that immutable retention cannot be bypassed by the same role, and that backup logs are written to a system the production identity cannot edit.
Decision rule: If the recovery copy can be modified by the same authority that runs production, escalate it as a trust-boundary failure and redesign the recovery path before relying on the backup for incident recovery.
Practitioner takeaway: A backup is only a recovery control when the attacker cannot rewrite the proof of recovery at the same time as the data itself.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on September 14, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org
