Back to Blog
2026-06-22· 6 min read

Novo Nordisk Hit by $25M Extortion — Hackers Stole AI Models and Clinical Trial Data

FulcrumSec extortion group stole 1.3TB from Novo Nordisk including clinical trial data and internal AI models. The hackers used AI agents to analyze the stolen models.

Ransomware Healthcare AI Security Extortion Pharmaceutical Data Theft

The Attack


Novo Nordisk, the Danish pharmaceutical giant behind Ozempic and Wegovy, confirmed a cybersecurity incident after the **FulcrumSec** extortion group claimed to have stolen **1.3 terabytes** of data including clinical trial information and internal AI models.


What Was Stolen


Clinical Trial Data

  • Pseudonymized patient IDs
  • Sex and year of birth
  • Biomarkers and health data
  • Immunogenicity data
  • Lifestyle factors (BMI, smoking status)

  • AI and Machine Learning Assets (Claimed)

  • 16.7 GB multimodal model checkpoint (text, image, transcriptomic data)
  • 407 MB proprietary biological/chemical training datasets
  • 50 MB source code for internal tool "NovoPert"
  • 113 training run logs
  • HPC infrastructure maps and Slurm configurations
  • 53 GB internal container images
  • Developer identities and private GitHub URLs

  • The Ransom Demand


    FulcrumSec demanded **$25 million** from Novo Nordisk. When the company refused to pay, the group began leaking data on June 15, 2026.


    AI vs AI: The Disturbing Innovation


    What makes this case unprecedented:


    1. **Attackers stole Novo Nordisk's AI models**

    2. **They ran their own AI agents** to analyze what they had

    3. **A second set of models** was used to adversarially critique the first analysis

    4. **The result**: A comprehensive report on the value of the stolen research


    FulcrumSec claimed this AI-generated analysis could save "other researchers or competitors 3-5 years of program development."


    The Entry Point: Stealer Logs


    Security researchers found that Novo Nordisk was heavily exposed in the infostealer economy:


  • 211 sets of employee credentials in stealer logs
  • 580 logins captured directly on Novo Nordisk pages
  • 2,932 session cookies harvested

  • This suggests the initial access may have come through credential theft via commodity infostealer malware.


    Impact on Healthcare


    For Pharmaceutical Companies

  • Protect AI/ML models as trade secrets
  • Implement zero-trust for research environments
  • Monitor for credential exposure in stealer logs
  • Segment AI infrastructure from general IT

  • For Healthcare Organizations

  • Clinical data is high-value target
  • Legacy systems need security attention
  • Third-party vendor risk is real
  • AI models represent new attack surface

  • How to Protect Your Organization


    1. **Monitor stealer logs** for leaked credentials (tools like Flare, SpyCloud)

    2. **Implement hardware security keys** for critical systems

    3. **Segment AI/ML infrastructure** from general network

    4. **Encrypt training data and models** at rest and in transit

    5. **Regular security assessments** of research environments


    Check Your Domain


    Scan your domain for vulnerabilities and exposed services.


    [Free security scan](https://vaarta.space)


    Ready to check your domain security?

    Run a free scan to identify potential vulnerabilities.

    Start Free Scan