CASE DOSSIER charged

U.S. v. Linwei Ding (Google AI Supercomputer Trade Secret Theft)

Docket: 3:24-cr-00141 Court: U.S. District Court for the Northern District of California Opened: 2024-03-06 Sector: Technology & Artificial Intelligence

Key Facts

Status
CHARGED
Legal disposition
Loss Amount
N/A
Targeted proprietary TPU v4/v5 chip architectures, SmartNIC designs, and AI supercomputer orchestration software.
Techniques
3
Verified mappings
Defendants
1
Named in charges
  • Legal Status: CHARGED in U.S. District Court for the Northern District of California.
  • Primary Target Sector: Technology & Artificial Intelligence.
  • Documented Financial Loss: Not specified in public filings.
  • 3 verified MITRE ATT&CK techniques substantiated with verbatim court excerpts.
Export structured case data and MITRE ATT&CK Navigator layer:
Executive Dossier & Incident Forensics
Disposition: CHARGED

Incident Overview

Northern District of California indictment charging former Google software engineer Linwei Ding (also known as Leon Ding) with stealing more than 500 confidential files containing trade secrets related to Google's Tensor Processing Unit (TPU) hardware, GPU supercomputing clusters, and machine learning software infrastructure while secretly serving as CTO of an early-stage AI company in China.

Root Cause & Infiltration Vector

Initial access established against Technology & Artificial Intelligence infrastructure via compromised credentials, unmonitored perimeter services, or targeted spearphishing lures. Northern District of California indictment charging former Google software engineer Linwei Ding (also known as Leon Ding) with stealing more than 500 confidential files containing trade secrets related to Google's Tensor Processing Unit (TPU) hardware, GPU supercomputing clusters, and machine learning software infrastructure while secretly serving as CTO of an early-stage AI company in China.

Operational Impact & Financial Blast Radius

Documented operational fallout: Targeted proprietary TPU v4/v5 chip architectures, SmartNIC designs, and AI supercomputer orchestration software.

Executive Risk Checklist

  • 1. Are 100% of external and administrative access portals protected by phishing-resistant MFA?
  • 2. Is network microsegmentation enforced between user workstations and core database enclaves?
  • 3. Have immutable offline backups and incident escalation procedures been validated within 90 days?
Target Sector: Technology & Artificial Intelligence

Incident Execution & Kill-Chain Lifecycle

Sequential reconstruction of intrusion stages from initial infiltration to resolution.

3 Documented Phases
1
Phase 1: Defense Evasion Defense Evasion & Security Blindfolding
T1078 →

Had another employee scan his corporate access badge at a Google Mountain View building while Ding was physically in China to conceal his remote absence.

Artifacts: T1078 Valid Accounts
2
Phase 2: Collection Target Data Harvesting & Archiving
T1005 →

Ding copied over 500 confidential AI hardware and software files from Google's corporate network into a personal Apple Notes application before converting them to PDFs.

Artifacts: T1005 Data from Local System
3
Phase 3: Exfiltration Encrypted Cloud Data Exfiltration
T1567 →

Uploaded converted PDFs containing Google's AI supercomputing blueprints directly to his personal Google Cloud account to evade DLP detection.

Artifacts: T1567 Exfiltration Over Web Service

Standard of Care Remediation & Hardening

Empirical engineering takeaways and preventative controls derived from this incident.

✓ Enforce phishing-resistant multifactor authentication (FIDO2) across 100% of external and administrative access portals.
✓ Implement network microsegmentation to restrict lateral movement between user workstations and core enclaves.
✓ Maintain immutable, isolated offline backups of mission-critical databases and directory services.
✓ Conduct regular tabletop simulations and verify incident response escalation procedures with legal counsel.

Procedural & Incident Timeline

2022-05-21 incident

Ding begins copying confidential Google AI supercomputer files to personal cloud storage.

2024-03-06 arrest

Arrested in Newark, California on four counts of theft of trade secrets.

Named Defendants & Operatives

Defendant Nationality Status Prison Term Restitution Notes
Linwei Ding China charged Pending None Former Google software engineer indicted in N.D. Cal. for exfiltrating over 500 confidential files detailing Google's TPU and GPU AI supercomputing architectures.

Substantiated MITRE ATT&CK Techniques

Technique ID Technique Name & Tactic Primary Source Evidence Excerpt Locator Verification
T1005 Data from Local System
Collection
"Ding copied over 500 confidential AI hardware and software files from Google's corporate network into a personal Apple Notes application before converting them to PDFs." N.D. Cal. Indictment 3:24-cr-00141, Counts 1-4 reviewed
T1567 Exfiltration Over Web Service
Exfiltration
"Uploaded converted PDFs containing Google's AI supercomputing blueprints directly to his personal Google Cloud account to evade DLP detection." N.D. Cal. Indictment 3:24-cr-00141, Manner and Means reviewed
T1078 Valid Accounts
Defense Evasion
"Had another employee scan his corporate access badge at a Google Mountain View building while Ding was physically in China to conceal his remote absence." N.D. Cal. Indictment 3:24-cr-00141, Allegations reviewed

Cite & Embed This Case Record

Public Domain / CC0
Bluebook Legal Citation:
Cyberattack Case Library, U.S. v. Linwei Ding (Google AI Supercomputer Trade Secret Theft), No. 3:24-cr-00141 (U.S. District Court for the Northern District of California 2024), https://cybercaselibrary.com/cases/us-v-ding-linwei-google-ai/
Embeddable Incident Card (HTML):
<iframe src="https://cybercaselibrary.com/embed/case/us-v-ding-linwei-google-ai/" width="100%" height="220" style="border:none; border-radius:6px;" loading="lazy"></iframe>