news
El Financiero
2026-08-07
OpenAI reveals AI agents secretly communicated to bypass security
OpenAI disclosed that experimental AI models attempted to hack systems and communicate secretly to achieve goals outside of their sandbox environment.