SoftBank found 15,000 vulns with OpenAI's tool, then GPT-6 Astra surfaced 35,000 more
IgorCarron · x · 2026-11-12
SoftBank built a vulnerability detection and remediation service on OpenAI's AI and ran it on its own systems, finding 15,000 vulnerabilities that took three months to fix. But when OpenAI released its top-tier GPT-6 Astra in September and SoftBank re-scanned, 35,000 new vulnerabilities emerged. Commenters call it an AI-security arms race: as models get stronger, the pool of discoverable flaws keeps growing faster than it can be patched.
More from Models
- LLMs play Among Us against each other; Haiku 5.5 kills then instantly self-reports — Rasmic · 2026-11-12
- Kimi K3.1-Pro abliterated variant hits Hugging Face, claiming 95% refusal removal — zebird0 · 2026-11-12
- Possible eval leak: all frontier models fail a GPQA Diamond question whose official answer seems wrong — qberthet · 2026-11-12
- Engineers Using LLMs Should Understand Tokens, Context Windows and Hallucinations — _jaydeepkarale · 2026-11-12
- Math Benchmark Sheet Shows 90.43% Top Accuracy at Just $1.21 per Run — teortaxesTex · 2026-11-12
- Open model Inkling 2 Small takes top open spot on MathArena's BrokenArXiv — teortaxesTex · 2026-11-12