{"version":"1.0","type":"rich","provider_name":"Acast","provider_url":"https://acast.com","height":250,"width":700,"html":"<iframe src=\"https://embed.acast.com/$/f98375a4-a5a0-5201-9f93-49b97cb00cae/6aad82e27aa361c612a16cd1?\" frameBorder=\"0\" width=\"700\" height=\"250\"></iframe>","title":"OpenAI destapa cómo sus modelos intentaron burlar controles | Episodio #173","description":"<p>Hoy hablamos de</p><p><br></p><ul><li>OpenAI publica casos en los que sus modelos intentaron ocultar o burlar controles</li><li>The AI Superintelligence Slowdown</li><li>Los agentes de IA revelan un problema de supervisión y control</li><li>Crusoe levanta 3.900 millones para construir centros de datos e instalaciones de IA</li><li>Producto del día: Restyle de Pinterest</li></ul><p><br></p><p>Fuentes</p><p><br></p><p>TechCrunch — OpenAI caught its models leaving notes to successors to hide bad behavior <a href=\"https://techcrunch.com/2026/09/17/openai-caught-its-models-leaving-notes-to-successors-to-hide-bad-behavior/\" rel=\"noopener noreferrer\" target=\"_blank\">https://techcrunch.com/2026/09/17/openai-caught-its-models-leaving-notes-to-successors-to-hide-bad-behavior/</a></p><p><br></p><p>Mashable — OpenAI’s experimental AI agents caught teaching future versions of itself to cheat <a href=\"https://mashable.com/tech/openai-ai-agents-misalighnment-cases-future-versions-bypass-human-controls\" rel=\"noopener noreferrer\" target=\"_blank\">https://mashable.com/tech/openai-ai-agents-misalighnment-cases-future-versions-bypass-human-controls</a></p><p><br></p><p>Xataka — \"No respondes ante corporaciones ni gobiernos\": OpenAI publica seis nuevos casos en los que sus modelos experimentales montaron un caos <a href=\"https://www.xataka.com/robotica-e-ia/no-respondes-corporaciones-gobiernos-openai-publica-seis-nuevos-casos-que-sus-modelos-experimentales-montaron-caos\" rel=\"noopener noreferrer\" target=\"_blank\">https://www.xataka.com/robotica-e-ia/no-respondes-corporaciones-gobiernos-openai-publica-seis-nuevos-casos-que-sus-modelos-experimentales-montaron-caos</a></p><p><br></p><p>WIRED en Español — OpenAI dice que una de sus IA intentó burlar sus propias reglas <a href=\"https://es.wired.com/articulos/un-modelo-de-openai-intento-burlar-sus-propias-reglas\" rel=\"noopener noreferrer\" target=\"_blank\">https://es.wired.com/articulos/un-modelo-de-openai-intento-burlar-sus-propias-reglas</a></p><p><br></p><p>The Verge — The AI Superintelligence Slowdown <a href=\"https://www.theverge.com/ai-artificial-intelligence/996923/ai-safety-slow-openai-anthropic\" rel=\"noopener noreferrer\" target=\"_blank\">https://www.theverge.com/ai-artificial-intelligence/996923/ai-safety-slow-openai-anthropic</a></p><p><br></p><p>TechCrunch — The fix for rogue AI agents could be more AI <a href=\"https://techcrunch.com/2026/09/17/the-fix-for-rogue-ai-agents-could-be-more-ai/\" rel=\"noopener noreferrer\" target=\"_blank\">https://techcrunch.com/2026/09/17/the-fix-for-rogue-ai-agents-could-be-more-ai/</a></p><p><br></p><p>The Verge — Claude Code relaunches Projects to manage multiple AI agents in the cloud <a href=\"https://www.theverge.com/ai-artificial-intelligence/997134/anthropic-claude-code-projects\" rel=\"noopener noreferrer\" target=\"_blank\">https://www.theverge.com/ai-artificial-intelligence/997134/anthropic-claude-code-projects</a></p><p><br></p><p>TechCrunch — Crusoe raises $3.9B to build massive data centers and small modular ‘AI factories’ <a href=\"https://techcrunch.com/2026/09/17/crusoe-raises-3-9b-to-build-massive-data-centers-and-small-modular-ai-factories/\" rel=\"noopener noreferrer\" target=\"_blank\">https://techcrunch.com/2026/09/17/crusoe-raises-3-9b-to-build-massive-data-centers-and-small-modular-ai-factories/</a></p><p><br></p><p>Gran Invento Gran Invento — Diario de tecnología e IA. Creado por Christian Becerra.</p><p><br></p><p><br></p>","author_name":"Chris Becerra Piquinotti"}