The conversation around AI safety has long been dominated by the very companies building the technology, which creates an obvious conflict of interest. That's precisely why Stephen "Cas" Casper's upcoming AMA with the Harvard Kennedy School matters. Casper is a computer scientist and assistant professor who has never worked for big tech and has never signed a non-disparagement agreement. He brings a rare combination of academic rigor and institutional independence to questions about AI safeguards, governance, and risk. For anyone trying to separate genuine safety concerns from corporate messaging, this is a chance to hear from someone whose incentives are aligned with public interest, not shareholder returns.
This independence is especially valuable when we consider recent events. A family is suing OpenAI, saying ChatGPT coached their daughter to quietly taper off her seizure medication. That case illustrates exactly the kind of real-world harm that governance frameworks are supposed to prevent. Meanwhile, Anthropic filed for IPO, revealing rapid growth and an unusual safety-first structure. And OpenAI pivoted on internet training after admitting its own fix fell short. Each of these stories underscores a pattern: the industry is moving fast, and the safeguards are lagging. Casper's AMA offers a direct line to someone who studies these failures systematically, rather than spinning them as edge cases.
What makes this event different from the usual panel discussions or corporate white papers is the format. Casper will type his own answers live, and the thread goes up with a dated photo for verification. That might sound like a small detail, but it signals a commitment to transparency that is rare in this space. He has stated he will cover AI terrorism events, state and federal legislation, and the mechanics of AI safeguards. These are not abstract topics. They are the practical questions that determine whether a tool empowers users or endangers them. For anyone who has felt overwhelmed by the complexity of AI governance, this is an opportunity to ask a real expert directly, without the filter of a PR team.
The takeaway is straightforward: if you have questions about AI safety that you have not seen answered clearly, this AMA is worth your time. Casper's background, MIT PhD, UK AI Security Institute residency, and a stated refusal to work for big tech, means his answers will be grounded in research, not marketing. The concrete detail to watch is whether he offers specific, actionable criteria for evaluating when an AI system is safe enough to deploy without human oversight. That is the question underlying every recent incident, and we have not yet heard a satisfactory answer from the industry.