Defenders Can Match 1,000 Attacking AI Agents—and Hold an Information Edge
marktenenholtz · x · 2026-11-08
- The debate: Quoting Calif CEO @XorNinja, who argues security has no playbook for stopping 1,000 AI agents attacking a network simultaneously: "attackers always adapt faster and earlier than defenders, even AI companies themselves don't have great security."
- The counterpoint: marktenenholtz argues the asymmetry can be neutralized—if attackers get 1,000 agents, defenders can field 1,000 too, and defender agents enjoy an information advantage.
- Google's mantra: The quoted thread notes Google accepts that prevention will fail at scale; security rests on prevention, detection, and response—likened to blindfolded cage fighting against an opponent who vanishes between punches.
- Takeaway: The core dispute is whether offense-defense asymmetry persists in the agent era, and whether detection + response can compensate for failed prevention.
Related event: Security CEO Warns Defenders Have No Playbook Against 1000 AI Agents(2 posts)→
More from AGI Musings
- Bindu Reddy pushes back on AI-peak narrative: GPU demand high, 99% of workflows unautomated — bindureddy · 2026-11-08
- Opinion: AI-built apps over YouTube and Amazon data turn platforms into gold mines — bendee983 · 2026-11-08
- Dev argues current models are already good enough—smarter AI only deepens uncertainty — JensHonack · 2026-11-08
- A year of deploying AI: where you deploy compute matters as much as how good it is — karinanguyen · 2026-11-08
- Richard Sutton: Brute force compute isn't enough, the science of AI is far from finished — RichardSSutton · 2026-11-08
- SimCity veteran: data centers should be every city's dream, so why the backlash? — signulll · 2026-11-08