---
title: "Securing AI agents: When AI tools move from reading to acting"
url: https://stacklist.com/card/54777167-ac54-4512-a58c-7041c26bfb56
source_url: "https://www.microsoft.com/en-us/security/blog/2026/06/30/securing-ai-agents-ai-tools-move-from-reading-acting/"
stack: https://stacklist.com/c/finance/stack/4f648364-9585-4a1c-8ecf-6de928fbc694
summary: "Microsoft Defender Experts observed increased ACR Stealer activity from late April 2026 to mid-June 2026, identifying two distinct intrusion chains across customer environments. The report also highlights the emerging risk of \"tool poisoning\" as AI agents evolve from passive reading to active execution capabilities."
tags: "acr-stealer, tool-poisoning, ai-agents, microsoft-defender, threat-activity, intrusion-chains, cybersecurity"
key_entities: "Microsoft (organization), ACR Stealer (technology), Microsoft Defender Experts (technology), Tool Poisoning (concept), AI Agents (concept), Intrusion Chains (concept)"
classification: "analysis"
content_hash: "sha256:0d6f7c3348f623a846fa8184531a7fbb49569d0ad20972d32e3898ea03b8e2d8"
acp_version: "0.2"
token_counts_approximate: 58
visibility: public
agent_accessible: true
status: "final"
---

# Securing AI agents: When AI tools move from reading to acting

July 16 12 min read ACR Stealer: Two observed intrusion chains amid increased threat activity From late April 2026 to mid-June 2026, Microsoft Defender Experts observed increased ACR Stealer activity across customer environments.