AI Security
AIDR cost drivers: how to budget without made up prices
What drives AI detection and response cost, which numbers to collect and what to ask vendors, so you can budget from your own volumes, not invented prices.
Oct 18, 20265 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
Zero AI, bring your own model or full AI: what teams get wrong
Six mistakes teams make choosing how a security tool uses models, from treating Zero AI as second class to skipping provider terms, plus a decision list.
Oct 18, 20264 min read
AI Security
Zero AI, bring your own model or full AI: how to choose
Choose the AI mode by what data the model sees, who runs it and what it may do. A checklist for tools that offer Zero AI, your own model or full AI.
Oct 18, 20264 min read
AI Security
Warehouse robots and AI agents: how the threat model changes
When AI agents direct warehouse robots, a software flaw can become a physical event. A sourced look at what changes in the threat model and where to start.
Oct 18, 20265 min read
AI Security
LLM10 unbounded consumption: a checklist for engineering leads
OWASP LLM10 covers runaway LLM cost, denial of service and model extraction. A checklist of limits, quotas, monitoring and access controls to put in place.
Oct 18, 20264 min read
AI Security
How to test for unbounded consumption in an LLM app (LLM10)
A defensive plan to test OWASP LLM10 unbounded consumption: limits on input, output, rate, cost and queues, plus alerts and graceful degradation.
Oct 18, 20264 min read
AI Security
How to test your LLM app for prompt injection (OWASP LLM01)
A defensive test plan for OWASP LLM01 prompt injection: map what the model can touch, test direct and indirect paths, and check least privilege.
Oct 18, 20265 min read
AI Security
How to test an LLM app for misinformation (OWASP LLM09)
OWASP LLM09 covers confident but false model output. Build a golden question set, check grounding, verify suggested packages and keep a human in the loop.
Oct 18, 20264 min read
AI Security
How do you test an LLM app for improper output handling?
Test LLM05 by listing every place model output lands, pushing harmless marker strings through each, and checking it is encoded or validated. Steps inside.
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.