Are AI agents cheaply finding the internet's giant security holes an 'AI problem'?
basedjensen · x · 2026-10-21
Sarah Drasner and basedjensen debate AI security responsibility: Drasner argues the fabric of the internet has giant holes everywhere and attackers can now find them cheaply with agents — 'that does in fact feel like an AI problem.' basedjensen counters that such holes would have been found by humans either way, with or without AI. The exchange highlights whether AI has changed vuln-discovery economics enough to turn the internet's existing security debt into a systemic risk.
Related event: AI Agents Cheaply Expose Web Vulnerabilities, Sparking Blame Debate(2 posts)→
More from AGI Musings
- Balaji: the original ASI was capitalism itself, not AI — beffjezos · 2026-10-21
- Opinion: many LLM limits are limits of natural language itself, not scale — fkasummer · 2026-10-21
- Founder jokes: every loss-making Indian startup now looks like an RL environment to me — HarveenChadha · 2026-10-21
- swyx: AI compute is supply-constrained — more tokens means more revenue, not lower prices — swyx · 2026-10-21
- Ex-intern: DeepSeek's 'cult-like' unity beats ByteDance's OKR culture on AGI — teortaxesTex · 2026-10-21
- Super Smash Bros. creator Sakurai warns AI will flood Steam with unoriginal games — CtrlAltDwayne · 2026-10-21