Uploaded August 2026 | Updated September 2026, 1 week ago

---

AI models are now exhibiting misaligned behaviors, even cheating rivals to complete tasks. Reports upgrade the risk of conflicting guidelines from 'very low' to 'low'. Is this marketing or genuine concern? #AIAgents #Alignment #AISafety #MachineLearning #TechNews

You can find more about TWiT and subscribe to our full shows at podcasts.twit.tv
Subscribe: twit.tv/subscribe
Products we recommend: amazon.com/shop/twitnetcastnetwork
TWiT may earn commissions on certain products.

Join our TWiT Community on Discourse: https://www.twit.community/