Exposed Cloud Training Apps Now a Major Enterprise Risk as Crypto-Mining and Exploits Spread

Training and demo web applications that are meant to run in safe, isolated environments are being found publicly exposed on enterprise cloud systems. Security research from Pentera Labs shows that these vulnerable applications are now being actively exploited by threat actors to gain access to broader cloud infrastructure, deploy crypto-mining software, and install persistence tooling across Fortune 500 organisations and leading security vendors.

Pentera Labs found nearly 2,000 exposed and vulnerable applications such as OWASP Juice Shop, Damn Vulnerable Web Application (DVWA), Hackazon, and bWAPP running on major cloud platforms including AWS, Azure, and Google Cloud Platform (GCP). Many of these appear on the public internet with minimal isolation, default configurations, and over-permissive cloud identity roles that give threat actors a foothold to expand their reach.

“One misconfigured training app was enough for attackers to obtain cloud credentials and deploy miners at an organization’s expense,” said Noam Yaffe, Senior Security Researcher at Pentera Labs and Team Lead of Pentera’s Offensive Security Services. “These systems may be labeled ‘non-production,’ but the access they expose is very real for thousands of organisations.”

These findings show that deployed training environments, when treated as low-priority assets, can quickly become entry points for attackers moving within an organisation’s cloud estate.

How Training Applications Are Being Misused

Many organisations spin up vulnerable demonstration applications for learning sessions, demos, or internal testing. These include open-source insecure projects intentionally built to demonstrate vulnerabilities.

In theory, these tools should run in isolated labs without real access to business systems. In practice, many are deployed inside live cloud accounts with broad identity and access management (IAM) roles and default credentials, creating a much larger attack surface than intended.

How Training Applications Are Being Misused

Many organisations spin up vulnerable demonstration applications for learning sessions, demos, or internal testing. These include open-source insecure projects intentionally built to demonstrate vulnerabilities.

In theory, these tools should run in isolated labs without real access to business systems. In practice, many are deployed inside live cloud accounts with broad identity and access management (IAM) roles and default credentials, creating a much larger attack surface than intended.

In verified cases, researchers identified 109 unique credential sets that could grant extensive permissions, including administrative access to cloud storage and compute resources.

Evidence of Active Exploitation

Pentera Labs documented real evidence of attacker abuse across these exposed instances.

About 20 percent of identified vulnerable applications showed clear signs of compromise. Malicious artifacts included:

• Crypto-mining software such as XMRig, actively mining Monero.
• Webshells and obfuscated scripts allowing command execution.
• Persistence mechanisms that survive attempted clean-ups.
• Credential theft and extraction scripts tied to overly permissive roles.

These active threats were not spread across small projects or isolated systems. Many were found on enterprise-owned cloud infrastructure, including instances connected with organisations like Cloudflare, F5, and Palo Alto Networks.

In some cases, vulnerable demos like DVWA were still using default credentials such as admin:password, which made exploitation trivial for automated tools and human attackers alike.

Why This Threat Is Dangerous for Cloud Security

Training and demo applications are often treated as temporary or “non-critical,” leading to three dangerous assumptions:

They will be removed or isolated later, so exposure is temporary.

They pose no real risk because they are not part of production systems.

Default configurations and credentials are acceptable because no real data lives there.

Pentera Labs’ findings challenge all three assumptions. Exposed training apps frequently remain online long after deployment. They often run alongside real workloads with cloud identities attached that have broad permissions. These identities let attackers pivot into more sensitive areas of the cloud environment.

Once attackers gain access to cloud metadata, they can move laterally, escalate privileges, or access services such as object storage, secrets managers, and container registries. They can even alter CI/CD pipelines or insert themselves into the organisation’s software supply chain.

Comparison: Training Apps vs. Production Risk

Category	Training App (Typical)	Production Environment
Intended Purpose	Security lab/demo/test	Business critical services
Access Controls	Often minimal	Usually enforced strict IAM
Exposure	Public internet	Typically internal / firewalled
Credential Risk	Linked to cloud identity roles	Managed securely with restrictions
Exploitation Evidence	High risk now confirmed	Major concern if breached

This comparison shows why training apps once thought isolated have become vectors for real attacks when left in production cloud accounts.

Best Practices to Mitigate Risk
To protect cloud environments from exploitation via training applications, organisations should:

To protect cloud environments from exploitation via training applications, organisations should:

Inventory all cloud assets including test, training, and demo apps.
Enforce least-privilege IAM roles and remove default credentials.
Isolate non-production environments from production cloud accounts.
Monitor and log access to exposed services for unusual activity.
Remove or relocate exposed training apps from public IPs.
Integrate continuous scanning and cloud posture checks in CI/CD pipelines.

FAQ: Cloud Training App Exposure and Exploitation