{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiczx4e7hlasdgcqo36csx5bumk5d36ojgjjpwo7wsqj4vdd5qnxi4",
    "uri": "at://did:plc:6wtxqaikjf62unmnajbfbq5v/app.bsky.feed.post/3mnku67zjh6i2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreictvlu7byth7evzyaznjqlrbyqmggqwlcvaerll5ez33lvj62sd7a"
    },
    "mimeType": "image/webp",
    "size": 58928
  },
  "path": "/microsoft-red-team-warns-agentic-ai-systems-can-bypass-human-approval-controls/",
  "publishedAt": "2026-06-05T17:34:32.000Z",
  "site": "https://vpncentral.com",
  "tags": [
    "News",
    "Microsoft Red Team Warns Agentic AI Systems Can Bypass Human Approval Controls",
    "VPN Central"
  ],
  "textContent": "Microsoft says a year of red teaming against deployed agentic AI systems revealed attack chains that can bypass human-in-the-loop controls and reach high-impact outcomes such as data exfiltration or lateral movement. The finding shows that approval prompts alone cannot protect AI agents that can plan, call tools, remember context, and act across multiple systems. The […]\n\nThe post Microsoft Red Team Warns Agentic AI Systems Can Bypass Human Approval Controls appeared first on VPN Central.",
  "title": "Microsoft Red Team Warns Agentic AI Systems Can Bypass Human Approval Controls"
}