AI Security
How to tier AI systems by risk: high, medium and low
Sort AI systems into risk tiers using data sensitivity, impact on people, autonomy and reach, so reviews match the real stakes of each system.
Oct 18, 20263 min read
Resources
Supply Chain Security, in plain English.
Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
Filter
Articles
RSS feedAI Security
How to test a public chatbot for wrong answers before launch
A pre launch test plan for a customer facing chatbot: gold answer sets, repeat runs, adversarial prompts and a takedown rule, drawn from reported failures.
Oct 18, 20263 min read
AI Security
How to set an AI risk tolerance your teams can actually apply
NIST's AI RMF does not prescribe risk tolerance. Here is a practical way to write one: levels, examples, limits, and who signs off on exceptions.
Oct 18, 20263 min read
AI Security
How to set a real code freeze for an AI coding agent
A code freeze for an AI agent must be enforced by permissions, not a prompt. Five controls to set up before the agent touches a shared system.
Oct 18, 20264 min read
AI Security
How to build an AI system inventory in one week
An AI system inventory is the base of AI governance. A one-week plan: scope, sources to scan, fields to record, an owner, and how to keep it current.
Oct 18, 20263 min read
AI Security
How much autonomy should an AI SOC agent have?
A decision guide to AI autonomy in security operations: from read-only summaries to supervised response, and why risky actions should wait for a person.
Oct 18, 20263 min read
AI Security
Hidden text in documents: protecting RAG ingestion from it
White-on-white text and other hidden content can carry instructions into a RAG index. How to extract text safely and validate documents before indexing.
Oct 18, 20264 min read
AI Security
Hidden text and invisible characters in LLM inputs: what to strip
Models read text people cannot see, like white-on-white resume text and invisible Unicode. How to extract, normalise and inspect input before the model.
Oct 18, 20263 min read
AI Security
Guard Proxy or Guard SDK: which fits your MCP setup?
Use the SDK when you own the MCP server code, the proxy when you do not. Both enforce the same policy. Components, trade offs and the current SDK status.
Oct 18, 20264 min read
Stay informed
Weekly insights on software supply chain security, delivered to your inbox.
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.