Attackers Exploit 87% of Vulns on Disclosure Day as GPT-6 Astra Aces ExploitBench
PeterDiamandis · x · 2026-09-14
Peter Diamandis published an essay on cybersecurity when both attackers and defenders run on AI:
- Attackers now exploit 87% of vulnerabilities on or before disclosure day, up from 23% in 2020; critical bug disclosures have jumped sixfold since spring — with AI driving both sides.
- GPT-6 Astra scored 100% on ExploitBench, writing working exploits for every known vulnerability; OpenAI rated it a "critical" cyber risk.
His argument: the model that can break into anything can also fix anything and rewrite 50 years of faulty human code. He frames a 12–24 month window to get on the right side of that, delivering the talk to 3,500 security professionals at GSX in Atlanta.
More from Models
- OpenAI CFO says 80% Luna price cut drove a 10X usage surge — Beth_Kindig · 2026-09-15
- Palantir, Nvidia, Booz Allen restrict Anthropic's Fable over data-retention concerns — dotey · 2026-09-15
- K2 Horizon 7B ranks between Qwen3.6 27B and 35BA3b on AA Intelligence Index — Uncle___Marty · 2026-09-15
- Post-training compute shift makes distillation of frontier models nearly impossible to prevent — maksym_andr · 2026-09-15
- Seattle Times and Newsday sue OpenAI and Microsoft over AI training on journalism — thione · 2026-09-15
- OpenBMB ships MiniCPM5-2B, a local open model billed as best in its size class — thione · 2026-09-15