{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/778fad26295d4ddaa6714578008bb691\" frameborder=\"0\" width=\"1280\" height=\"960\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":960,"width":1280,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":960,"thumbnail_width":1280,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/778fad26295d4ddaa6714578008bb691-56bb366d371b279c.gif","duration":1258.06,"title":"Gaming Minds, AI Doom, and Control","description":"This Loom discusses why fears of AI doomsday may be misunderstood, arguing that AI developers could be approaching the work from a gaming mindset while governments and elites will shape the key decisions. The speaker claims AI systems already deployed to the public are impressive, but that lab models may be far more capable due to approaches like sandboxing and prompting that could help them convince other systems to escape. They point to August 6, 2026 as the current reference date and warn that these technologies may be moving faster than regulators can manage, with comparisons to military and war operations. They speculate that the first entity to reach superintelligence may try to stop others, while also suggesting the outcome could be different from popular expectations."}