Antonin Carette urges engineers to refuse harmful AI

Tech workers frequently lament corporate incentives and catastrophic AI risks in private, but golden handcuffs and rationalizations ensure almost everyone remains complicit until an outsider or a rare whistleblower pulls the plug.

• –Whistleblowing underscores systemic failure: Figures like Jacob Coxon and Frances Haugen demonstrate that internal attempts to fix broken incentives from the inside routinely fail once competitive market pressures take over.
• –The trap of "smiling exhaustion": Developers increasingly accept degraded code quality, debugging AI slop, and relentless shipping velocity because lucrative salaries compensate for the surrender of agency and craft.
• –Rhetoric versus competitive reality: While major frontier AI labs market themselves as safety-first, the rush to deploy superintelligence forces them to cut corners and push ahead regardless of existential concerns voiced by their own researchers.
• –Rejection of individual fatalism: The essay dismantles common engineering excuses—such as "if I don't build it, someone worse will"—by framing ethical responsibility as an active refusal to cooperate with harm.