Unit 42 — 2026 IR report on AI-accelerated attacks

AI relevance: Unit 42 finds AI is now compressing the attack lifecycle (recon, phishing, scripting, execution), shrinking time-to-exfiltration for real-world intrusions.

  • Unit 42 analyzed 750+ major incidents across 50+ countries to identify 2026 attack trends.
  • AI is cited as a force multiplier that makes attacker operations faster and more scalable.
  • In the fastest cases, attackers moved from initial access to exfiltration in ~72 minutes, a 4× speedup.
  • Identity weaknesses featured in nearly 90% of investigations, reinforcing “log in, don’t break in.”
  • Attackers leveraged third‑party SaaS integrations in 23% of cases, expanding supply‑chain risk.
  • Multi‑surface intrusions dominated: 87% involved activity across endpoints, cloud, SaaS, and identity.
  • The browser was involved in 48% of incidents, reflecting how routine workflows are now attack vectors.
  • Extortion is shifting away from encryption toward data‑theft‑first pressure tactics.

Why it matters

  • AI‑accelerated tradecraft shrinks defender reaction time, so slow detection becomes a breach amplifier.
  • AI ops teams must treat identity and SaaS integrations as primary attack surfaces, not secondary risks.

What to do

  • Harden identity: tighten MFA, session lifetimes, token scopes, and privileged access paths.
  • Instrument AI pipelines with detection and response automation that can keep pace with sub‑hour attacks.
  • Audit SaaS integrations and apply least‑privilege to connectors used by AI/agent workflows.
  • Prioritize browser security (isolation, protections, telemetry) given its growing role in intrusions.

Sources