It Will Kill Us No company should get to decide how much danger the rest of us have to live with without telling us what’s at stake. Even if its products are useful. Even if millions of people choose them. Even if the people building them believe they are improving the world. When human lives are at stake, good intentions are not a safety strategy. We should be asking whether these systems are aligned with human wellbeing. Whether their creators are honest about the risks. And whether their incentives reward making our lives better or simply making their products harder to live without. Scrutiny should be as widespread as adoption, with particular attention to incentives. These companies promise to make our lives easier. They get rewarded for making their products more compelling, expanding distribution and increasing use. But widespread adoption is not proof of widespread benefit. Public safety has to be an obligation. We can’t run a population-scale experiment in which the financial rewards are private and the risks belong to everyone. I’m talking about ultra-processed food, by the way. If you thought I was talking about AI, there are plenty of parallels with the current debate. And as with food, I’m optimistic about its future. I believe in AI’s potential benefits, and I want the people building it held accountable for its risks. I also want that same standard applied to my industry. A 2025 CDC report found that ultra-processed foods supplied roughly 62% of daily calories for American children and teenagers. Separately, research in adults has linked higher consumption to greater risks of type 2 diabetes and early death. At sweetgreen, our job is to make real food so delicious, convenient and accessible that people choose it. The fear surrounding AI is that we’ll build something enormously profitable, become dependent on it and discover too late that human wellbeing was never a condition of its success. The food industry doesn’t have to imagine that future. We have to stop accepting it as the present.