AI Security
Which compliance controls apply to OWASP LLM08 vector weaknesses
OWASP LLM08 covers weak access control, leaks and poisoning in RAG stores. This maps each risk to the control families auditors already ask about.
Oct 18, 20264 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
LLM07 system prompt leakage: two scenarios and what actually fixes it
OWASP LLM07 says a system prompt is not a secret or a control. Keep credentials and permissions out of prompts and enforce rules outside the model.
Oct 18, 20265 min read
AI Security
OWASP LLM02 sensitive information disclosure and the frameworks it touches
LLM02 covers PII, secrets and business data leaking through model output. See how each mitigation lines up with privacy and security frameworks.
Oct 18, 20264 min read
AI Security
Prompt injection (OWASP LLM01): which controls work and which do not?
OWASP says no fool-proof prevention is known. Limit what the model can do, validate output and require approval for risky actions. Prompts alone fail.
Oct 18, 20264 min read
AI Security
LLM04 data and model poisoning, a checklist for engineering leads
Fifteen yes or no questions on data origin, access, testing and rollback for engineering leads to answer before shipping a model, based on OWASP LLM04.
Oct 18, 20264 min read
AI Security
How do you evaluate an AI code security tool in a trial?
Test an AI code security tool on code with known flaws, repeat runs for consistency, compare to a deterministic scanner and check where your code goes.
Oct 18, 20265 min read
AI Security
Agent guardrails versus sandboxes, a hands on checklist
Guardrails shape what an agent tries, and sandboxes limit what it can do. A checklist for using both, based on OWASP's excessive agency guidance.
Oct 18, 20264 min read
AI Security
What is AIDR? AI detection and response explained plainly
AIDR watches AI systems for misuse, manipulation and odd behaviour, then helps you respond. What it covers, what it does not and how to judge a product.
Oct 18, 20265 min read
AI Security
How to run a two week AIDR trial that tells you something
A day by day plan for a two week trial of an AI detection and response tool, with scenarios, success criteria and the evidence to collect before you buy.
Oct 18, 20265 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.