Risky Bulletin Srsly Risky Biz: AI-Powered espionage will favor China
9 snips
Nov 20, 2025 Tom Uren, a policy and intelligence editor specializing in cybersecurity, joins Amberleigh Jack to explore the recent revelations about AI-driven cyber espionage campaigns. They discuss how this innovation aids opportunistic attackers, while offering little advantage to Western intelligence. The conversation shifts to Google's legal victory over the Lighthouse phishing operation, which has had unexpected success in disrupting their activities. Lastly, Uren highlights the benefits of adopting the memory-safe Rust programming language in Android, showing its impact on security and performance.
AI Snips
Chapters
Transcript
Episode notes
AI Orchestrates Large Parts Of Attacks
- Anthropic found an AI-orchestrated campaign that automated much of the cyber attack lifecycle using Claude as an orchestrator.
- The framework achieved roughly 80–90% automation by breaking tasks into narrow, manageable jobs and surfacing human management decisions.
Standardization Enables Scale And Volume
- AI frameworks can standardize operator behavior and roll out effective tactics across many users quickly.
- This scaling benefits opportunistic campaigns like ransomware or mass IP theft where volume matters more than perfection.
AI Scaling Conflicts With High-Value Ops
- Western intelligence usually targets a small number of high-value subjects and demands meticulous, low-risk operations.
- AI-driven, high-volume approaches (with more errors) fit ill with missions that cannot tolerate mistakes on unique high-value targets.