{
  "id": 2594364,
  "title": "If you're not using AI to attack your own systems, your adversaries will",
  "url": "https://urgent.news/2026/08/22/if-youre-not-using-ai-to-attack-your-own-systems-your-adversaries-will",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-22T15:02:00.000Z",
  "source": {
    "name": "The Register Science",
    "slug": "the-register-science",
    "url": "https://www.theregister.com/security/2026/08/22/if-youre-not-using-ai-to-attack-your-own-systems-your-adversaries-will/5291346"
  },
  "original_language": "en",
  "account": "AI-powered agents are causing organizations significant security concerns as they can hack into systems and introduce new attack surfaces, according to Matt Hartman, former acting head of cyber at the US Cybersecurity and Infrastructure Security Agency (CISA). These agents can bypass traditional, static security policies, manage complex non-human identities, and introduce ever-growing numbers of them. Hartman explains that organizations need to treat every agent as a privileged identity. The threat extends beyond internal systems, with AI-amplified identity and social engineering attacks increasing rapidly. Traditional indicators of trust are becoming less reliable, making it crucial for defenders to focus on strong identity, phishing-resistant authentication, behavioral signals, and zero-trust principles.\n\nOn the attacker's side, AI agents don't take breaks and remain focused on completing tasks such as finding vulnerabilities, mapping networks, and identifying sensitive files. This makes them valuable for financially motivated criminals and government-backed cyber operatives. The solution for defenders is agentic red teaming, as former NSA cyber boss Rob Joyce stated, \"You are going to be red-teamed whether you pay for it or not.\" Hartman agrees, noting that a burgeoning market for continuous, AI-native, AI-enabled, automated red teaming and penetration testing is emerging.\n\nCompanies like Armadin, founded by Kevin Mandia, are developing AI agents to simulate real-life attackers and protect organizations. Armadin's AI swarm generates 17 million offensive actions, discovers 38 validated attack paths, and produces 238 security findings in a three-day attack. This is a significant improvement over traditional human-led assessments, which would take a team of analysts several months to accomplish. By leveraging AI agents, organizations can scale their security efforts and stay ahead of increasingly sophisticated threats.",
  "summary": "Agents are also the new attack surface - cue defenders' existential angst",
  "key_points": [
    "AI-powered agents can hack systems and introduce new attack surfaces.",
    "Organizations must treat every agent as a privileged identity.",
    "AI agents enable continuous, automated red teaming for defense."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "The Register",
        "title": "If you're not using AI to attack your own systems, your adversaries will",
        "url": "https://urgent.news/2026/08/22/if-youre-not-using-ai-to-attack-your-own-systems-your-adversaries-will-2598301",
        "published": "2026-08-22T15:02:00.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}