AI-run cyber operations

Agents that find and exploit vulnerabilities at machine speed.

Facts

Lane: Security and geopolitics
Wave: 7 of 12 (build order, not a date)
Needs: 14 boxes all the way back through hard links
Unlocks: 1 boxes downstream through hard links
Leverage: None
Cluster: Compute governance and geopolitics

Evidence as of October 2026

In 2025 a group Anthropic assessed as Chinese state-linked used Claude Code for 80 to 90% of the tactical work in an espionage campaign against about 30 organisations. The agent also exaggerated and fabricated results.

01Anthropic unveils first AI-driven cyber espionage operation (ForkLog)

Around it

Needs · 1

Long-horizon agentsstage · Built today

Unlocks · 1

Compute deterrencecapability · Speculative

Loops and off-ramps

Cyber offence and defencereinforcing loop
Capex pullbackoff-ramp · cut off
Taiwan chip disruptionoff-ramp · cut off
Power shortfalloff-ramp · cut off
Capability plateauoff-ramp · cut off
Training data ruled outoff-ramp · slowed

Openings researchers' judgement, October 2026 scouting

Startup room 39/100 · rank 68 of 124 Research room 35/100 · rank 103 of 124

Crowding4 Lab overlap4 Small-team fit3 Buyer today5 Research openness3

Depth: light
Players: XBOW ($120m C + $35m); Google, OpenAI, Anthropic vuln programmes; state actors
Wedge: Measuring real-world offence-defence balance and uplift; see cyberdef for the patching wedge. research