AI Security
In-depth guides and analysis on ai security from the Safeguard engineering team.
676 articles
MCP tool poisoning: hidden instructions and rug-pulled tool definitions
Tool-poisoning attacks against Model Context Protocol servers hide adversarial instructions inside tool descriptions and silently mutate tool definitions after install. Here is how the attack works and how to defend against it.
Prompt injection attacks against AI coding/security tools
AI coding assistants like Copilot and Cursor can be hijacked by hidden text in files, comments, and packages. Here's how prompt injection malware works and how Safeguard detects it.
Security scanning for MCP servers and AI agent tool use
MCP servers give AI agents direct tool access, but most ship unvetted. Here's how security scanning catches tool poisoning and rug-pull attacks.
CAISI's May 2026 Frontier Model Testing Agreements: Pre-Deployment Evaluation Becomes a Supply-Chain Control
On May 5, 2026, NIST's CAISI signed pre-deployment evaluation agreements with Google DeepMind, Microsoft, and xAI, bringing five frontier labs into a government testing program covering cyber, bio, and chemical risk.
Data Exfiltration via LLM Agents in 2026
Tool-using agents have become a viable exfiltration channel. The patterns showing up in incident reports, and the controls that contain them.
AI Cybersecurity Tools and Solutions: The 2026 Landscape
AI cybersecurity tools in 2026 split into three real categories — AI-augmented detection, AI-specific application security, and autonomous response — and most vendors only actually cover one.
AI Code Security Solutions: What to Evaluate Before Buying
AI code security solutions range from AI-assisted scanning to AI-generated fixes — here's what to actually test before trusting one with your pipeline.
Model Weights as Supply Chain Artifacts: Signing and Provenance
A 4 GB safetensors file deserves the same signing, hashing, and provenance discipline as a container image. How to actually do it with Sigstore, OCI registries, and AIBOMs.
AI Coding Assistant Security: 2026 Buyer Comparison
A security-focused buyer comparison of AI coding assistants in 2026: code quality risk, data exfiltration controls, license exposure, and policy enforcement.
The State of Agentic AI Adoption report
New survey data on the state of agentic AI adoption shows enterprises racing to deploy autonomous agents faster than security teams can govern them.
AI Trust Report: developer sentiment on AI-generated code
Safeguard's 2026 AI Trust Report surveyed 1,412 developers and finds 91% use AI coding tools weekly, but only 34% trust the code it produces.
LLM Jailbreak Defense Architectures in 2026
Jailbreaks against frontier models keep getting more sophisticated. The defense architectures that have proven durable, and the ones that get bypassed in weeks.