Metadata
| Source Table | entities |
| Source ID | tool-restrictions |
| Entity Type | approach |
| Description | Tool-use restrictions limit what actions and APIs AI systems can access, directly constraining their potential for harm. This approach is critical for agentic AI systems, providing hard limits on capabilities regardless of model intentions, with METR evaluations showing agentic task completion horiz… |
| Wiki ID | E487 |
| Children | 1 total(1 fact) |
| Created | Mar 29, 2026, 10:41 PM |
| Updated | Oct 8, 2026, 1:34 AM |
| Synced | Oct 8, 2026, 1:34 AM |
Record Data
id | tool-restrictions |
wikiId | E487 |
stableId | Tool-Use Restrictions(approach) |
entityType | approach |
title | Tool-Use Restrictions |
description | Tool-use restrictions limit what actions and APIs AI systems can access, directly constraining their potential for harm. This approach is critical for agentic AI systems, providing hard limits on capabilities regardless of model intentions, with METR evaluations showing agentic task completion horiz… |
website | — |
tags | [ "agent-safety", "capability-restrictions", "defense-in-depth", "deployment-safety", "permission-systems", "mcp-security" ] |
clusters | [ "ai-safety", "governance", "cyber" ] |
status | — |
lastUpdated | 2026-02 |
customFields | — |
relatedEntries | [
{
"id": "sandboxing",
"type": "approach"
},
{
"id": "sid_95ASV0XEwQ",
"type": "capability"
},
{
"id": "sid_rQEkFJxS5w",
"type": "organization"
},
{
"id": "sid_mK9pX3rQ7n",
"type": "organization"
},
{
"id": "sid_1LcLlMGLbw",
"type": "organiz… |
metadata | — |
Debug info
Thing ID: sid_nl9oD17gfg
Source Table: entities
Source ID: tool-restrictions
Wiki ID: E487
Entity Type: approach