Editorial companion

What stands out

Gary Marcus, co-founder of Robust.AI and a long-time skeptic of large-scale artificial intelligence (AI), tells Romaine Bostick that major developers, including OpenAI, have not put in adequate safeguards for systems they cannot reliably control. The blunt framing, that some deployed models are effectively uncontrollable, stands out because it comes from a practitioner who has worked on both commercial and research fronts. The clip crystallizes a central debate: whether rapid product rollouts outpaced engineering and governance practices needed to manage real-world risk.