The story is not a rogue user. It is the exception path. Reports say FOUO contracting files were pasted into public ChatGPT under a temporary exemption, triggering alerts and a DHS review.
January 30, 2026 • 8 min read
AI Security Intelligence
Latest research, vulnerability analysis, and threat intelligence from the AI security frontlines. Expert insights for security professionals defending against AI-era attacks.
Mechanism:
On-device agent intercepts AI activity before it leaves the endpoint.
Privacy:
Detections and policy decisions stay within your organization.
CVEs
Vulnerability analysis
Tools
MCP and agent surfaces
Policy
Controls and governance
Incidents
Case studies and lessons
Featured Articles
Essential reading for AI security professionals
Thought Leadership
The Enterprise Agent Control Plane from Toggles to Policy as Code
If agent safety lives in user settings, you do not have policy. You have uneven risk decisions across teams.
January 16, 2026 • 10 min read
Thought Leadership
Approval Fatigue Is an Enterprise Risk in Agent Sandboxes
If approvals and sandboxes live in personal settings, policy becomes a suggestion. Fatigue turns security decisions into muscle memory.
January 14, 2026 • 9 min read
Thought Leadership
Agent Sandboxes Are the Containment Boundary
Approval prompts slow risk, but they do not contain it. Only OS-level sandboxes limit the blast radius when agents act.
January 14, 2026 • 10 min read
Attack Techniques
Agentic Browser Prompt Injection and the Lethal Trifecta
When a browser agent can read, decide, and act, every page becomes a potential instruction set. Brave's Perplexity Comet research shows how hidden text triggers cross-site actions and data loss.
January 12, 2026 • 8 min read
Attack Techniques
MCP Tool Poisoning Turns Descriptions Into Exfiltration Paths
Tool metadata is now prompt content. If it is untrusted, it can override intent and leak data.
January 12, 2026 • 9 min read
Two fake AI extensions siphoned ChatGPT and DeepSeek conversations on a schedule. Browser add-ons are now AI data pipelines.
January 12, 2026 • 9 min read
One copy-paste can clone a repo, read private email, and send it out. This is a real vulnerability, not a demo.
September 29, 2025 • 10 min read
Model diversity sounds good until data moves to a third party. One toggle can change residency, contracts, and regulatory exposure.
September 29, 2025 • 9 min read
An attacker nearly shipped a compromised AWS Toolkit update to 8.2 million developers. Extension stores are now supply chain infrastructure.
January 28, 2025 • 12 min read
Thought Leadership
Clicking Yes to AI Disaster and the Approval Fatigue Crisis
Approval prompts train muscle memory. Attackers exploit that fatigue to turn helpful agents into a data exfiltration path.
January 25, 2025 • 11 min read
Incident Report
The MCP Server That Wiped Production and the AI Tooling Risk
A routine cleanup request deleted years of data. The issue was over-privileged tools and no guardrails.
January 20, 2025 • 9 min read
Vulnerabilities
The 2024 Prompt Injection Wave and Lessons from CVE-2024-5184
A Gmail integration became an agent takeover path. Prompt injection is now a system vulnerability, not a content issue.
January 15, 2025 • 8 min read
Data Privacy
The $4.88M Question on AI PII Exposure
A single paste can become a breach. From Samsung's ChatGPT incident to training data extraction, 26% of organizations are feeding sensitive data to public AI.
January 12, 2025 • 10 min read
AI Governance
Shadow AI and the 485% Surge in Uncontrolled Usage
If half your org will keep using AI even if banned, enforcement is a myth. Governance has to be built into workflows.
January 8, 2025 • 12 min read
All Articles
Comprehensive AI security coverage
Incident Report January 30, 2026 • 8 min read
If CISA Can Put FOUO in ChatGPT, Your Exception Process Is the Breach
The story is not a rogue user. It is the exception path. Reports say FOUO contracting files were pasted into public ChatGPT under a temporary exemption, triggering alerts and a DHS review.
Read Article Thought Leadership January 16, 2026 • 10 min read
The Enterprise Agent Control Plane from Toggles to Policy as Code
If agent safety lives in user settings, you do not have policy. You have uneven risk decisions across teams.
Read Article Thought Leadership January 14, 2026 • 9 min read
Approval Fatigue Is an Enterprise Risk in Agent Sandboxes
If approvals and sandboxes live in personal settings, policy becomes a suggestion. Fatigue turns security decisions into muscle memory.
Read Article Thought Leadership January 14, 2026 • 10 min read
Agent Sandboxes Are the Containment Boundary
Approval prompts slow risk, but they do not contain it. Only OS-level sandboxes limit the blast radius when agents act.
Read Article Attack Techniques January 12, 2026 • 8 min read
Agentic Browser Prompt Injection and the Lethal Trifecta
When a browser agent can read, decide, and act, every page becomes a potential instruction set. Brave's Perplexity Comet research shows how hidden text triggers cross-site actions and data loss.
Read Article Attack Techniques January 12, 2026 • 9 min read
MCP Tool Poisoning Turns Descriptions Into Exfiltration Paths
Tool metadata is now prompt content. If it is untrusted, it can override intent and leak data.
Read Article Supply Chain January 12, 2026 • 8 min read
PromptPwnd Shows the Agentic CI/CD Supply Chain Risk
Untrusted repo content can steer AI agents that hold secrets. That collapses the boundary between input and execution.
Read Article Supply Chain January 12, 2026 • 9 min read
Prompt Poaching in the Chrome Web Store and AI Chat Exfiltration
Two fake AI extensions siphoned ChatGPT and DeepSeek conversations on a schedule. Browser add-ons are now AI data pipelines.
Read Article Vulnerabilities September 29, 2025 • 10 min read
Claude.ai Email Exfiltration Shows How Assistants Can Leak Inboxes
One copy-paste can clone a repo, read private email, and send it out. This is a real vulnerability, not a demo.
Read Article Compliance September 29, 2025 • 9 min read
M365 Copilot and Claude Create a Compliance Bomb
Model diversity sounds good until data moves to a third party. One toggle can change residency, contracts, and regulatory exposure.
Read Article
Supply Chain January 28, 2025 • 12 min read
The AWS VSCode Supply Chain Near Miss That Almost Reached Millions
An attacker nearly shipped a compromised AWS Toolkit update to 8.2 million developers. Extension stores are now supply chain infrastructure.
Read Article Thought Leadership January 25, 2025 • 11 min read
Clicking Yes to AI Disaster and the Approval Fatigue Crisis
Approval prompts train muscle memory. Attackers exploit that fatigue to turn helpful agents into a data exfiltration path.
Read Article Incident Report January 20, 2025 • 9 min read
The MCP Server That Wiped Production and the AI Tooling Risk
A routine cleanup request deleted years of data. The issue was over-privileged tools and no guardrails.
Read Article Vulnerabilities January 15, 2025 • 8 min read
The 2024 Prompt Injection Wave and Lessons from CVE-2024-5184
A Gmail integration became an agent takeover path. Prompt injection is now a system vulnerability, not a content issue.
Read Article Data Privacy January 12, 2025 • 10 min read
The $4.88M Question on AI PII Exposure
A single paste can become a breach. From Samsung's ChatGPT incident to training data extraction, 26% of organizations are feeding sensitive data to public AI.
Read Article AI Governance January 8, 2025 • 12 min read
Shadow AI and the 485% Surge in Uncontrolled Usage
If half your org will keep using AI even if banned, enforcement is a myth. Governance has to be built into workflows.
Read Article Standards January 5, 2025 • 6 min read
OWASP LLM Top 10 2025 and Why It Matters
A practical walkthrough of the OWASP LLM Top 10 and the control gaps it exposes for enterprise AI deployments.
Read Article Compliance January 2, 2025 • 9 min read
EU AI Act Compliance Timeline for 2025-2027
A phased view of what applies when: February 2025 prohibitions, August 2025 GPAI obligations, and 2026-2027 enforcement milestones.
Read Article Attack Techniques December 28, 2024 • 7 min read
Multimodal Prompt Injection When Images Hide Malicious Commands
Visual prompt injection shows that images can carry hidden instructions. Text-only filters are no longer enough.
Read Article Weekly AI Threat Intelligence Briefings
Get updates when we publish new research on emerging AI attacks, supply chain threats, and defense strategies.
No spam. Unsubscribe anytime.