Hijack SageMaker notebook via lifecycle config
Can attach a startup script to a notebook and run code as its execution role.
Grants escalation only in combination. All of these are needed together:
sagemaker:CreateNotebookInstanceLifecycleConfigsagemaker:UpdateNotebookInstancesagemaker:StopNotebookInstancesagemaker:StartNotebookInstancesagemaker:UpdateNotebookInstance can attach a startup script to a notebook and run code as its execution role, but only when it is held together with sagemaker:CreateNotebookInstanceLifecycleConfig, sagemaker:StopNotebookInstance, sagemaker:StartNotebookInstance. Held together they reach administrator access.
Why it is dangerous
A principal can create a notebook lifecycle configuration with an attacker-controlled startup script, stop an existing notebook that runs as the target role, attach the config with sagemaker:UpdateNotebookInstance, and start the notebook again. On startup the script executes as the target role and reads the target role's live credentials, no iam:PassRole needed, because the target role is already bound to the notebook. If the target role is privileged this is full privilege escalation against a notebook that was already trusted.
In MITRE ATT&CK terms this maps to T1578, Modify Cloud Compute Infrastructure, under the privilege escalation tactic. AWS records it in CloudTrail as CreateNotebookInstanceLifecycleConfig, UpdateNotebookInstance, StopNotebookInstance, StartNotebookInstance. The call is loud in CloudTrail and stands out from normal activity.
Detection
CreateNotebookInstanceLifecycleConfigUpdateNotebookInstanceStopNotebookInstanceStartNotebookInstancePrerequisites
An existing SageMaker notebook instance with an execution role; permission to create a lifecycle config and to stop/update/start the notebook.
MITRE ATT&CK
T1578 · Modify Cloud Compute InfrastructureTactic
Privilege Escalation
A control that would remove it: Restrict sagemaker:UpdateNotebookInstance / lifecycle-config actions on notebooks that carry a privileged execution role.
Whether to apply it depends on how the permission is used in your environment, so it is a business decision rather than an automatic fix. The same permission can be load-bearing in one account and pure risk in another.
Pathbreak confirms whether this path resolves in your account, ranks it against the others by how much it opens up, and uses your threat model and business context to show which paths are worth cutting.
Questions
What does sagemaker:UpdateNotebookInstance allow?
Can attach a startup script to a notebook and run code as its execution role.
What else is needed for sagemaker:UpdateNotebookInstance to be dangerous?
It becomes an escalation only when held together with sagemaker:CreateNotebookInstanceLifecycleConfig, sagemaker:StopNotebookInstance, sagemaker:StartNotebookInstance. On its own it is not enough, which is what makes the combination easy to miss.
How is hijack SageMaker notebook via lifecycle config detected?
Watch for CreateNotebookInstanceLifecycleConfig, UpdateNotebookInstance, StopNotebookInstance, StartNotebookInstance in CloudTrail. The call is loud in CloudTrail and stands out from normal activity.
How do you break this path?
A control that would remove it is: Restrict sagemaker:UpdateNotebookInstance / lifecycle-config actions on notebooks that carry a privileged execution role. Whether to apply it depends on how the permission is used in your account, so it is a business decision rather than an automatic fix.
Which MITRE ATT&CK technique is this?
T1578, Modify Cloud Compute Infrastructure, under the privilege escalation tactic.
Related paths
References
Find this in your own account
We run a read-only scan of one AWS account and show you which of these paths exist, ranked by blast radius, with the change that breaks each one. First validated paths within the hour.