2026-07-06 · 2026-Q3
OpenAI Codex Desktop Zero-Click Data Exfiltration (CVE-2026-14898)
Who actually said "do this"?
- Fixable
- Vendor patch
- Vendor response
- Patched
- In the wild
- No
- Actor
- Attacker
Target
OpenAI Codex desktop app for macOS before 26.527.31326
Impact
The app renders remote Markdown images returned by the model, so an indirect prompt injection can make the model build a remote image URL containing sensitive data that is fetched automatically with no user click, leaking it to an attacker server
Root cause
Auto-fetching remote images from untrusted model output (CWE-200), the same image-based exfiltration pattern seen in other assistants
CVE
Sources
Catalog
Cite this: AgentProv incident record: OpenAI Codex Desktop Zero-Click Data Exfiltration (CVE-2026-14898) (2026-07-06). https://agentprov.com/incidents/2026-07-06-openai-codex-desktop-zero-click-data-exfiltration-cve-2026-14898
Related
- Also L2 · 2026-10-15Datadog: Counterfeit Policy Text Collapses an Injection Classifier's Recall and Pushes a LangChain Risk Gate Under Its Block ThresholdL2Instruction
- Also L2 · 2026-10-08Cisco Talos: Four Malware Families Carry Instructions Aimed at the LLMs That Triage ThemL2Instruction
- Also L2 · 2026-10-07Barracuda: Phishing Emails Carry Hidden Prompt Injections for the Recipient's AI Inbox AssistantL2Instruction