Disclosed vulnerability
Amazon Q for Business: markdown hyperlink rendering enabled data exfiltration
AIR-2024-0001 AIR record · 2024-01-amazon-q-business-markdown-exfil
From the supporting source
An Indirect Prompt Injection attack can cause the LLM to return markdown tags.embracethered.com
The record
Disclosed
2024-01-18
Vendor
Amazon Web Services
Product
Amazon Q for Business
Risk domain
Privacy › Data exfiltration channels
Also touches
Human manipulation, Data and memory poisoning, Confused deputy
System type
Generative model
Who acted
Agent acted — Amazon Q rendered the exfiltration hyperlink
Attack surface
Enterprise assistant
Autonomy level
Assistant
Initial vector
Indirect prompt injection document
Impact type
Data exfiltration, Phishing
Confidence
High
Submitted by
EAI Internal
Crosswalk
- OWASP Agentic Top 10 (2026)
- ASI02 – Tool Misuse and Exploitation
- MITRE ATLAS
- AML.T0024 – Exfiltration via AI Inference API
- NIST AI RMF
- MAP 4.2
- EU AI Act
- Article 10
- ISO/IEC
- TR 24028; 23894
- AIUC-1
- A. Data & Privacy
- Scope
- tool misuse
- Mitigation
- Monitor network traffic for anomalous patterns; Restrict egress destinations.
- Taxonomy
- Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
Amazon Q for Business tenants
Capability abused
Defense & remediation
Recorded defense
none
Guardrail kinds
None
Guardrail outcome
Not applicable
Fix shipped
fixed and deployed worldwide 2023-12-29, about a week after reporting; URLs replaced with a placeholder
Taxonomy & references
Curator notes
Johann Rehberger.
Supporting sources
Cite as
Enkrypt AI. "Amazon Q for Business: markdown hyperlink rendering enabled data exfiltration". AIR, AIR-2024-0001. https://agentincident.vercel.app/i/AIR-2024-0001/ (accessed 2026-10-19).