Incident Score: Analysis & Impact (GOO1781699328)

Rankiteo's analysis has identified several MITRE ATT&CK tactics and techniques associated with this incident, each with varying levels of confidence based on available evidence. Under the Initial Access tactic, the analysis identified Supply Chain Compromise: Compromise Software Dependencies and Development Tools (T1195.002) with moderate to high confidence (80%), supported by evidence indicating flaw in Google Cloud’s Vertex AI Python SDK (google-cloud-aiplatform) and Exploit Public-Facing Application (T1190) with moderate to high confidence (70%), supported by evidence indicating vertex AI’s Model Registry relies on GCS buckets to stage artifacts. Under the Resource Development tactic, the analysis identified Acquire Infrastructure: Cloud Accounts (T1583.001) with high confidence (90%), supported by evidence indicating attacker pre-created buckets in their own project (bucket squatting) and Compromise Infrastructure: Cloud Storage (T1584.005) with high confidence (90%), supported by evidence indicating bucket squatting technique to hijack staging buckets. Under the Execution tactic, the analysis identified Exploitation for Client Execution (T1203) with moderate to high confidence (80%), supported by evidence indicating unsafe Python pickle deserialization to execute arbitrary code and Content Injection (T1659) with high confidence (90%), supported by evidence indicating replaced legitimate model files with malicious payloads within 2.5s. Under the Persistence tactic, the analysis identified Account Manipulation: Additional Cloud Roles (T1098.003) with moderate to high confidence (70%), supported by evidence indicating configured permissive IAM roles on attacker-controlled buckets. Under the Privilege Escalation tactic, the analysis identified Domain Policy Modification: Group Policy Modification (T1484.001) with moderate confidence (60%), supported by evidence indicating oAuth token exfiltration from Google-managed service accounts and Exploitation for Privilege Escalation (T1068) with moderate to high confidence (80%), supported by evidence indicating rCE via pickle deserialization granted broad cloud-platform scope. Under the Defense Evasion tactic, the analysis identified Modify Cloud Compute Infrastructure: Create Snapshot (T1578.001) with moderate to high confidence (70%), supported by evidence indicating malicious model treated as legitimate due to absent integrity checks and Masquerading: Match Legitimate Name or Location (T1036.005) with high confidence (90%), supported by evidence indicating predictable bucket naming allowed bucket squatting in attacker’s project. Under the Credential Access tactic, the analysis identified Steal Application Access Token (T1528) with high confidence (90%), supported by evidence indicating oAuth token exfiltration from Google-managed service accounts. Under the Discovery tactic, the analysis identified Cloud Service Discovery (T1526) with moderate to high confidence (80%), supported by evidence indicating access to BigQuery metadata and internal infrastructure data and Account Discovery: Cloud Account (T1087.004) with moderate to high confidence (70%), supported by evidence indicating cross-deployment data access and reconnaissance capabilities. Under the Collection tactic, the analysis identified Data from Cloud Storage (T1213.003) with high confidence (90%), supported by evidence indicating model theft and access to BigQuery metadata. Under the Exfiltration tactic, the analysis identified Transfer Data to Cloud Account (T1537) with moderate to high confidence (80%), supported by evidence indicating oAuth tokens and model artifacts exfiltrated via RCE. Under the Impact tactic, the analysis identified Data Manipulation: Stored Data Manipulation (T1565.001) with high confidence (90%), supported by evidence indicating poisoned ML model artifacts deployed as legitimate and Resource Hijacking (T1496) with moderate to high confidence (80%), supported by evidence indicating cross-tenant RCE undermined tenant isolation in Vertex AI. These correlations help security teams understand the attack chain and develop appropriate defensive measures based on the observed tactics and techniques.