Critical Docker AI Assistant Flaw Enabled Remote Code Execution via Metadata
Docker patched a severe vulnerability in Ask Gordon AI (CVE-2026-24578) allowing code execution through malicious image metadata. Learn the technical details and mitigation steps.
Docker Patches Critical Vulnerability in AI Assistant Ask Gordon
Cybersecurity researchers at Noma Labs have disclosed details of a now-patched critical vulnerability in Ask Gordon, an AI assistant integrated into Docker Desktop and the Docker Command-Line Interface (CLI). The flaw, tracked as CVE-2026-24578 (codename: DockerDash), could be exploited to achieve remote code execution (RCE) and data exfiltration via maliciously crafted image metadata.
Technical Details of CVE-2026-24578
The vulnerability stemmed from improper input validation in Ask Gordon’s metadata processing engine. Attackers could embed malicious payloads within Docker image metadata, which the AI assistant would then execute when parsing the image. Key attack vectors included:
- Arbitrary code execution via crafted metadata fields
- Privilege escalation within Docker environments
- Sensitive data exposure through exfiltration channels
Noma Labs researchers demonstrated that the flaw could be triggered without user interaction, making it particularly severe for developers and DevOps teams relying on Docker’s AI-assisted workflows.
Impact Analysis
The vulnerability posed significant risks to organizations using Docker Desktop or CLI with Ask Gordon enabled:
- Unauthorized system access via RCE
- Supply chain attacks through compromised container images
- Data leakage of credentials, API keys, or proprietary code
Given Docker’s widespread adoption in cloud-native development, the flaw had the potential to impact thousands of enterprises and open-source projects.
Mitigation and Recommendations
Docker has released patches addressing CVE-2026-24578 in the following versions:
- Docker Desktop 4.32.0 (or later)
- Docker CLI 26.1.0 (or later)
Security teams should:
- Immediately update Docker Desktop and CLI to the latest versions
- Audit container images for suspicious metadata entries
- Disable Ask Gordon in high-security environments until patches are applied
- Monitor for unusual activity in Docker logs, particularly metadata parsing events
Noma Labs has published a technical deep dive with proof-of-concept details for security professionals.
Original report: The Hacker News