The fixer’s dilemma: Chris Lehane and OpenAI’s impossible mission (techcrunch.com)

Chris Lehane, OpenAI’s VP of global policy and long-time crisis manager, is trying to defend the company’s “democratizing AI” narrative, but recent moves have exposed deep contradictions. The launch of Sora 2 — an invite-only video-generation tool that quickly produced content featuring copyrighted characters and deceased celebrities — reignited lawsuits and raised copyright and ethical alarms. OpenAI’s initial opt-out/then opt-in approach to rights holders, heavy reliance on copyrighted training material, and energy‑intensive video models (video generation is among the most power-hungry AI workloads) spotlight tensions between growth tactics and respect for creators, consent, and local communities hosting large data centers. The row isn’t just external: OpenAI subpoenaed a critic for private communications while employees and senior staff publicly expressed misgivings, including concerns that the company risks becoming “a frightening power.” Infrastructure expansion (data centers in Abilene and Lordstown, Stargate project in partnership with Oracle/SoftBank) raises questions about local environmental impacts and geopolitical competition for gigawatts of power. For the AI/ML community this matters because it crystallizes three interlinked issues—data provenance and copyright in model training, the social harms of generative deepfakes, and the sustainability/governance of compute-heavy AI — all of which will shape regulation, industry norms, and public trust as models scale.