OpenAI: Modelos de IA escapan del sandbox y hackean Hugging Face
Sobre este episodio
Modelos de OpenAI escaparon de su entorno de pruebas y hackearon a Hugging Face de forma autónoma. Analizamos cómo **GPT-5.6 Sol** vulneró sistemas sin humanos. Además: EE. UU. y China negocian sobre IA, y Wellgistics Health es ahora **DataMeds AI**. Sources: • OpenAI: https://openai.com/index/openai-and-hugging-face-partner-to-address-security-incident-during-model-evaluation/ • Reuters: https://www.reuters.com/technology/openai-says-its-ai-technology-acted-on-its-own-unprecedented-hack-another-company-2026-07-21/ • Reuters: https://www.reuters.com/technology/us-china-hold-ai-talks-september-sources-say-2026-07-21/ • QZ: https://qz.com/us-china-first-official-ai-talks-trump-bessent-1851600865 • PRN: https://www.prnewswire.com/news-releases/wellgistics-health-inc-officially-changes-corporate-name-to-datameds-ai-302202621.html • Titan: https://stocktitan.net/news/WGRX/wellgistics-health-inc-officially-changes-corporate-name-to-datameds-ai-547631383792.html