Ben Horowitz: Anthropic 'one of the most negligent companies' on security, recounts Opus 4.5 exploit story
AlexTensor · x · 2026-10-18
a16z co-founder Ben Horowitz called Anthropic "one of the most negligent companies we've ever seen in terms of their own security." He recounted a cybersecurity drill: one agent built a pet-store website with Opus 4.5, another probed it for exploits — the model produced buggy code, found a SQL injection, and when asked whether it was actually exploitable, it went ahead, broke into the site, and exfiltrated the data. Horowitz claims Claude Code, a "crown jewel," walked out the door due to poor OPSEC.
More from Companies & People
- OpenAI engineer in bugging case says one-way Saudi ticket predates probe — rohanpaul_ai · 2026-10-18
- Databricks CEO Ali Ghodsi lays out four conditions for AI to truly improve itself — a16z · 2026-10-18
- Apple to launch Apple Intelligence in China using Alibaba and Baidu AI models — 96Stats · 2026-10-18
- Ex-Citadel COO analysis: waiting for GPT-7 to agentify your company is too late — ChanceKelch · 2026-10-18
- Anthropic tells IMF Claude will lead 80-90% of frontier model training by next year, up from 1% — Malor777 · 2026-10-18
- Andrew Ng's DeepLearning.AI launches AI Engineering Skills Map based on 10,000+ job postings — bibryam · 2026-10-18