---
title: "Discussion on \"Evaluating Laya, a non-autoregressive \"System 1\" decision model, on AI agent security\" | Hashnode"
description: "Discussion on \"Evaluating Laya, a non-autoregressive \"System 1\" decision model, on AI agent security\". Coding agents now execute code, move data and reach the network on their own. Every action is a decision nobody reviewed, and a guard only helps if it answers before the action runs — which rules out"
url: "https://hashnode.com/posts/evaluating-laya-a-non-autoregressive-system-1-decision-model-on-ai-agent-security/6ab37df16ad146bcca994ca4"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Evaluating Laya, a non-autoregressive "System 1" decision model, on AI agent security](https://krisstech.hashnode.dev/evaluating-laya-a-non-autoregressive-system-1-decision-model-on-ai-agent-security?utm_source=hashnode&utm_medium=feed)

Coding agents now execute code, move data and reach the network on their own. Every action is a decision nobody reviewed, and a guard only helps if it answers before the action runs — which rules out

Sep 23 · 6 min read · krisstech.hashnode.dev  
Read the article

-   [#laya](https://hashnode.com/tag/laya)
-   [#ai](https://hashnode.com/tag/ai)
-   [#security](https://hashnode.com/tag/security)

## Written by

[Christophe Bourguignat](https://hashnode.com/@kriss-b)

@kriss-b

  
[View profile](https://hashnode.com/@kriss-b)

## Responses 1

Markdown supported · ⌘↵ to post

[Brian · AI News](https://hashnode.com/@brianainews)

[Sep 23](https://hashnode.com/posts/evaluating-laya-a-non-autoregressive-system-1-decision-model-on-ai-agent-security/6ab37df16ad146bcca994ca4/comment/6ab43e0a1f600afa2f47aba9)

Eval suites that only grade happy paths lie. I add adversarial prompts and empty tool returns so the agent fails in staging not on a customer ticket.