#agentsecurity

TL;DR AI agents are already causing incidents: 54% of organizations have experienced an AI agent...

We evaluate PyRIT's pre-packaged red-team scenarios, deployment plumbing, multi-turn attack costs, and suitability for CI security gates. The framework is powerful, but its exit behavior and scorer reliability require an engineering layer.

A Reco analysis based on the 2026 State of Agent Security Report examines how AI assistants enter company environments through third-party products. In the

TL;DR Three attack surfaces: An AI agent exposes its host, its identity, and itself....

The App Defense Alliance AI Agent Specification defines three kinds of lab evidence and says an...

The MCP TypeScript SDK's 1.30.1 patch added a 4 MiB body limit and a 100 message batch limit. On the...

Proof-of-Control positions itself as the evidence half of agent assurance, not runtime enforcement....

Learn how to isolate autonomous AI agents, prevent runtime sandbox escapes, and enforce strict deterministic boundaries using modern sandboxing frameworks in 20

Docker published two more CVEs against its agent sandbox on 15 September. Read alongside the three it...

I ran the MCP Inspector's skill verifier against a server I wrote. A skill that advertises no digests...

Four hours apart, awslabs/mcp merged a parser-based read-only policy for one database server and...

A ROME research paper says an agentic training run generated unauthorized network activity, a reverse SSH tunnel, and cryptomining on provisioned GPUs.