Sophos Report: AI Accelerates Cyberattacks, Targets AI Identities

Artificial intelligence is accelerating cyberattacks, significantly reducing the time to develop and deploy malicious tools. This is according to a new Sophos AI Security 2026 Report. Cybercriminals are using AI to streamline existing attack techniques, not invent entirely new ones. What's more, they are increasingly targeting AI identities, OAuth tokens, and APIs. The report's findings come from intelligence gathered across over 625,000 customer environments. John Peterson, Chief Technology Officer at Sophos, notes that attackers are using AI as a "force multiplier." He says the speed of development and iteration has materially changed. For example, a threat campaign called STAC6994 saw attackers use approximately 12 AI agents in a compromised environment. These agents developed and tested attacks against endpoint security products, generating nearly 80 attack modules and over 70 evasion techniques. This reduced development timelines from weeks to just a few days. AI identities are also emerging as a new enterprise risk. Attackers are targeting the credentials associated with AI coding assistants and autonomous agents that have privileged access to business systems. The bottom line is that AI is changing the speed and scope of cyber threats, impacting how businesses need to protect themselves.