AI Security
How Do You Secure Generative AI When Its Outputs Vary?
Generative AI answers the same prompt differently, so one passing test proves little. Use NIST and OWASP risk lists and enforce limits outside the model.
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
How do secrets end up in prompts and context windows?
Secrets reach prompts through pasted logs, env files, tool output and retrieved documents, then persist in transcripts. Reduce exposure, scan inputs and rotate quickly.
Oct 18, 20264 min read
AI Security
How do you sandbox an AI agent that runs shell commands?
Run agent commands in a disposable container or microVM with no secrets, a read-only base, limited network and resource caps, and a human gate on anything that leaves it.
Oct 18, 20264 min read
AI Security
How do you run an MCP server in Docker safely?
Docker can sandbox a local MCP server that would otherwise run with your privileges. Use stdin, a read-only root, no capabilities and limited network.
Oct 18, 20264 min read
AI Security
How do you run an LLM locally with Docker Model Runner?
Docker Model Runner pulls open models and serves them on an OpenAI-compatible API, keeping prompts on your machine. Setup basics and the gaps to close.
Oct 18, 20264 min read
AI Security
How do you run a private mirror for Hugging Face models?
Stage vetted models internally, control the cache location, and go offline in production. Here is the setup and what a mirror does not verify.
Oct 18, 20264 min read
AI Security
What role does AI play in endpoint security, and how can it be evaded?
Machine learning classifies malware and odd behavior on devices, but NIST documents poisoning and evasion attacks on such models. See how to use it safely.
Oct 18, 20265 min read
AI Security
What are the risks and benefits of AI in application security?
AI speeds code review, triage and fixes but adds new risks, from wrong answers to excess agency. Weigh both and keep humans and tests in the loop.
Oct 18, 20265 min read
AI Security
What does responsible AI usage mean in practice for a company?
Responsible AI means clear ownership, tested systems and honest limits. See the OECD principles, the NIST AI RMF functions and steps a team can start with.
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.