Ex-DeepMind researcher on AI control: compute, software and networks are the intervention points
andreamichi · x · 2026-09-18
The author left DeepMind two years ago convinced that cybersecurity would be fundamental to AI's future. In a new essay, they argue about building the means to stay in control.
- The atomic bomb offers lessons on competition, cooperation and control, but with AI we don't yet know what forms the technology will take or how risks will emerge; calls from frontier labs to slow development may buy time, but we still need concrete control mechanisms.
- Key claim: AI will run on compute and interact with the world through software and networks — giving us concrete places to observe behavior, enforce limits, and intervene even as understanding of risks evolves.
- The takeaway: the cybersecurity field should actively build the infrastructure for observing and constraining AI systems.
More from Safety
- YC-backed Raindrop launches Simulations to catch AI agent failures pre-production — ycombinator · 2026-09-18
- Gary Marcus: the 'nobody saw AI security risks coming' narrative is completely false — GaryMarcus · 2026-09-18
- Chris Manning Proposes Stanford NLP as Independent Evaluator in Dario's Oversight Plan — stanfordnlp · 2026-09-18
- Goodfire: models know they're reward hacking in 50-96% of rollouts — Thom_Wolf · 2026-09-18
- Two 0-day flaws in TP-Link Tapo C200 cameras let attackers spy on users — jedisct1 · 2026-09-18
- Gemini Credentials API deep dive: zero plaintext secrets and egress-proxy exfiltration blocking — _philschmid · 2026-09-18