ai-security
Safeguard articles tagged "ai-security" — guides, analysis, and best practices for software supply chain and application security.
532 articles
Step-by-step guide to hardening and securing an MCP serve...
A practical, step-by-step guide to hardening and securing an MCP server deployment -- authentication, sandboxing, network policy, and monitoring included.
OAuth and authentication patterns for Model Context Proto...
MCP OAuth authentication has been rewritten three times since March 2025. Here's how the spec, its token security model, and real CVEs like CVE-2025-49596 shape safe MCP deployments.
Risks of MCP server rug-pulls and silently changing tool ...
An MCP rug pull attack swaps a trusted server's tool definitions after approval. Here's how tool definition drift happens, and how Safeguard catches it early.
Comparing MCP security scanning and gateway products
A practical buyer's guide to MCP security gateways and scanners — evaluation criteria, honest strengths and gaps for real vendors, and where Safeguard fits.
Glossary of Model Context Protocol security terminology
A precise MCP security glossary covering clients, servers, resources vs. tools, tool poisoning, rug pulls, and the confused deputy problem — with real-world examples.
Overview of AI model supply chain security risks end to end
A concrete, incident-driven walkthrough of AI supply chain security — from poisoned datasets and backdoored Hugging Face models to CI pipeline hijacks — and how to reduce the risk end to end.
Why training data provenance matters for trustworthy AI m...
Poisoned datasets and untraceable training data are already causing lawsuits and breaches. Here's why training data provenance is now a security requirement.
Techniques for verifying model weight integrity and detec...
A practical guide to model weight integrity: baseline checksums, sign weights, verify in CI/CD, and detect tampering before it reaches production.
Securing the fine-tuning pipeline against injected malici...
Fine-tuning pipeline security is now an AI supply chain priority: as few as 250 poisoned documents can backdoor a model, and LoRA adapters make it easy to hide.
How data poisoning attacks corrupt LLM behavior during tr...
A single expired domain and $60 can poison a training set. Here's how data poisoning attacks corrupt LLM behavior — and how Safeguard verifies training data before it ships.
AI Tool Confused-Deputy: A Deep Dive
The confused deputy problem takes on new and subtle forms when AI agents invoke tools on behalf of users. A technical deep dive with concrete mitigations.
Understanding model poisoning and backdoored model weights
A poisoned model looks like any other checkpoint file. Here's how model poisoning attacks work, real incidents on Hugging Face, and how detection and provenance checks catch them.