These two talks are unrelated, but are interesting conversations about AI Safety from people outside the AI Safety Community.
Gebru: Eugenics and the Promise of Utopia through Artificial Intelliegence
Most of the talk is critical analysis of EA and adjacent communities, although at 37 mins she pivots to discussing visions of AGI, and at about 47 minutes discusses why AGI is inherently unsafe. This last section on why AGI is inherently unsafe I think will get a lot of agreement from people here.
Lazar: Generative AI and the New Bing
The talk is much less focused on AI XRisk specifically, although a lot of it is likely relevant for people thinking. The section around 34 minutes- 39 minutes on the , whilst short, is explicitly relevant to EA concerns