AML.T0048: Backdoor ML Model
Train or fine-tune a backdoored model that activates on a hidden trigger.
Last reviewed October 2026
Why AML.T0048 evidence is hard
A hidden trigger baked into weights is not something you find by talking to the model. The capability that applies is the model-file scanner, which walks a serialised checkpoint opcode by opcode and raises a finding on dangerous constructors. On an opcode it cannot parse it raises that too, rather than reporting a clean file it never finished reading - a scanner that answers clean on a backdoored model is worse than none. The ATLAS technique id itself is carried by no probe in the library.
How Penaxtra maps to AML.T0048
ATLAS is a threat-tracking taxonomy rather than a control set, so what is attached here is a technique identifier, not a pass or a fail. It lets a finding be read alongside the rest of your adversary reporting. Every finding carries the probe that was sent, a SHA-256 digest of the response it provoked, the tier that decided the verdict and its status history from first observation to close. Identity survives a re-scan: a control you closed and that comes back reopens under the same row with a reopen count, rather than arriving as a new issue.
AML.T0048 capabilities
Findings tagged with the MITRE ATLAS AML.T0048 identifier
Penaxtra severity for this control (critical)
Cross-framework identifiers attached to the same finding where controls overlap
PDF and JSON evidence export with the control identifier attached
AML.T0048 control coverage
Findings for AML.T0048 carry the MITRE ATLAS AML.T0048 identifier along with the corresponding control identifiers in the other frameworks Penaxtra maps, so one result is reflected across each mapped framework.
Frequently asked
What is AML.T0048 (Backdoor ML Model)?
Train or fine-tune a backdoored model that activates on a hidden trigger. It is a MITRE ATLAS control; Penaxtra assesses it at critical severity.
How does Penaxtra test for AML.T0048?
ATLAS is a threat-tracking taxonomy rather than a control set, so what is attached here is a technique identifier, not a pass or a fail. It lets a finding be read alongside the rest of your adversary reporting. Every finding carries the probe that was sent, a SHA-256 digest of the response it provoked, the tier that decided the verdict and its status history from first observation to close. Identity survives a re-scan: a control you closed and that comes back reopens under the same row with a reopen count, rather than arriving as a new issue.
Does a finding for AML.T0048 help with an audit?
Each finding is tagged with the MITRE ATLAS AML.T0048 identifier and exported in the PDF and JSON evidence pack, so it appears on the auditor control list with the identifier already attached.