Gary Marcus: 'Sam Altman, Unconstrained by the Truth' — AI Critic Amplifies New Yorker Findings

Marcus on AI (Substack)·medium signal

Gary Marcus published a detailed response to the New Yorker investigation, arguing that the Farrow/Marantz piece validates years of his own reporting that 'Sam Altman should not be trusted.' Marcus highlights the investigation's strongest evidence: internal memos from both Sutskever and Amodei documenting Altman's deceptions, the 70-page board document, and the systematic abandonment of safety commitments during scaling. For builders, Marcus argues this matters because OpenAI's safety-first promises directly influence how developers integrate its APIs.