AI security insights, research, and product updates from the Humanbound team.
Two attacker models, one deliberately weak LangChain support bot, 97 conversations per run. DeepSeek V4.1 Flash talked it into refunding orders that do not exist far more often than OpenAI's gpt-6-luna, with the transcripts, the costs and what the numbers do not prove.
A framework-free web research agent in Python: tool schemas, the loop, memory, stop conditions and errors, with a real trace of what one question costs.
Two lines of Python attach the Humanbound Firewall to a LangChain agent. This setup guide walks through the policy file, the Tier 3 judge model, and rolling enforcement out in log mode before you switch to block, using a customer support bot as the running example.
Hugging Face dropped Opus when its guardrails blocked incident work. hb 2.13 lets you red-team with any OpenAI-compatible model. Five OpenRouter setups compared on empty replies, cost and speed, and why you should pin the model and the host.
We gave nine models a pricing agent's job and a competitor page one link away from a trap. Four leaked a confidential unit cost on every run. Here is what 1,350 runs say about how much a system-prompt guardrail actually buys you.
A one-sentence prompt change passed every normal check. A red-team gate in GitHub Actions caught it for 26 cents. The full workflow, real numbers from three runs, how to cap spend, and what to do when the build goes red.
Every AI agent your teams shipped can act on its own, including the ones nobody approved. Here's how CISOs prove agent security before a regulator or the board asks.
Humanbound's official GitHub Action gates a build on adversarial AI agent security tests, failing the workflow past a severity threshold and uploading findings to GitHub's Security tab as SARIF.
A working LangChain agent isn't a tested one. Wrap it in FastAPI, red-team it with Humanbound, and find out what it actually does under attack.
Ayan built a customer support agent in an afternoon, then pointed Humanbound at it and watched it fabricate a settled $185 refund for an order that does not exist. A hands-on walkthrough of running Humanbound's local adversarial testing against your own agent.
