Anthropic has warned that artificial intelligence is making cyberattacks larger and more complex. The warning comes in its September threat intelligence report.

The report covers cases the company disrupted between December 2025 and August 2026. Anthropic says it blocked AI misuse linked to cyberattacks, surveillance and disinformation campaigns worldwide.

One key finding: attackers are moving from using AI for simple assistance toward letting AI orchestrate whole attack chains. This means fewer human steps and faster attacks.

The report also describes how AI speeds up surveillance and the spread of false information. Automated systems can now do in hours what once took teams of people.

These are vendor-reported findings, based on what Anthropic saw on its own systems. Independent experts say the trend matches what the wider security industry is seeing.

Cybersecurity agencies advise companies to patch software, use strong passwords and train staff. Experts say defence must also use AI, since manual responses are too slow for automated attacks.