Data Exfiltration
Vendor AI Connector Data Flows and Exfiltration Risk
How forgotten OAuth tokens and unmonitored AI connectors become data exfiltration highways.
Marcus Oduya
· 9 min read
AI Threat Bulletin — Page 2
Data Exfiltration
RAG Knowledge Base Exposure Through Adversarial Queries
RAG systems leak sensitive data through a structural flaw, not model safety issues.
Derek Tso10 min read
Data Exfiltration
Data Exfiltration via LLM Output Channels in Agentic Workflows
Agents with hands turn prompt injection into unauthorized data theft at scale.
Solène Marchetti10 min read
LLM Vulnerabilities
Confidential Computing Controls for LLM Inference in Enterprise Environments
Hardware-enforced isolation protects sensitive data during cloud-based LLM inference.
Priya Venkataraman10 min read
LLM Vulnerabilities
Insecure Output Handling Vulnerabilities in LLM-Integrated Applications
Treat all LLM-generated output as untrusted input before it reaches downstream systems.
Solène Marchetti10 min read
LLM Vulnerabilities
OWASP LLM Top 10 Coverage Gaps for Enterprise Deployments
The framework misses enterprise risks in agentic AI, identity governance, and attack chaining.
Priya Venkataraman10 min read
Vendor AI Risk
Continuous Monitoring of Vendor AI Assets After Deployment
Vendors silently update AI models in production, leaving deployers exposed to undetected drift.
Marcus Oduya9 min read
AI Agent Security
Persistent Threat Techniques in Long-Running AI Agents
Corrupted memory in AI agents persists across sessions, turning exploits into standing instructions.
Derek Tso10 min read
AI Agent Security
Credential and Token Theft via Compromised AI Agents
Attackers chain three mechanisms to steal tokens from AI agents.
Solène Marchetti11 min read
AI Agent Security
Agentic AI Risk Challenges Beyond Traditional TPRM Controls
Agentic AI systems shift risk profiles between reviews, exposing gaps in annual vendor assessments.
Marcus Oduya11 min read
AI Agent Security
Red Teaming OpenAI and Anthropic Agent Builder Platforms
Agents that take real actions demand red teaming methods fundamentally different from chatbots.
Derek Tso10 min read
AI Agent Security
Supply Chain Risk in AI Agent Tool Registries
Malicious skills poisoned AI agent registries before defenses could catch up.
Priya Venkataraman8 min read
