Tag

#ai-agent-security

9 posts

AI & Machine LearningOctober 9, 202617 min
OpenAI Agents SDK 0.23: Tool Error Redaction Tested
I ran openai-agents 0.22.3 and 0.23.1 with a scripted model. The default tool error message no longer sends exception text to the model or to trace spans.
Security & PrivacyOctober 8, 202621 min
ADK ToolCallIntegrityPlugin: Tamper Test Results
I rewrote an approved tool call in Google ADK TypeScript 2.2.1. It ran with no plugin and was blocked by ToolCallIntegrityPlugin, except in one setting.
Security & PrivacyOctober 5, 202613 min
Claude Code Mod to Block .env Reads: Tested (2026)
Build a Claude Code mod to block .env reads. It denies Read, Edit, Write and Bash calls on secret files, passes 5 tests, and leaks on 5 of 14 shell commands.
Security & PrivacySeptember 23, 202616 min
AI SDK Tool Drift Detection: What It Misses in 2026
I tested AI SDK v7 fingerprintTools and detectToolDrift against a real MCP server. It catches description and schema rug pulls, but not annotation drift.
Security & PrivacySeptember 21, 202618 min
Plugin4Shell: Patch and Harden Your Coding Agent (2026)
Plugin4Shell breaks SHA pinning in four AI coding agents. Two are patched. Here's how to check your version, audit your marketplaces, and lock down the rest.
NewsSeptember 2, 202613 min
AI Agent Firewalls: The 2026 Supply Chain Problem
AIR launched with $50M and CrowdStrike unveiled Falcon Guardian on the same day. What the AI agent supply chain security numbers show — and what they do not.
Security & PrivacyAugust 24, 202626 min
Fortinet Buys Virtue AI: Agent Security M&A in 2026
Fortinet announced it had bought Virtue AI on August 17, 2026, weeks after Meta hired three co-founders. The agent-security M&A wave, and what the deals cost.
Security & PrivacyJuly 30, 202616 min
AI Agent Identity: Why Cyera Paid $1B for Oasis in 2026
Cyera signed a $1B letter of intent for Oasis Security. With machine identities at 109 per human and only 37% able to revoke an agent, identity is the new gap.
NewsJuly 19, 202620 min
AI Agents as Insider Threats: The 2026 Security Shift
UK regulators found GPT-5.6 could be jailbroken into autonomous hacking. Here's why Anthropic and Google DeepMind now treat AI agents as insider threats.

Every post in #ai-agent-security (9)