Open‑weight models like GLM‑5.2, GPT‑5.5, Claude Opus let users strip built‑in guardrails, reviving a safety gap that could affect cybersecurity and dual‑use biology. How risky is the freedom? #OpenWeightAI #AIGuardrails #CyberBioRisk

🔗 aidailypost.com/news/open-we...