AI reflects the incentives, risks, and blind spots of the institutions that build and deploy it — but users still have agency. Some guardrails are needed.
• high-confidence falsehoods,
• harm-enabling instructions,
• defamation risk,
• medical / legal liability,
• political persuasion at scale
Open AI has considerable investment in ethics which is necessary for risk management in a mass produced LLM but also important for the nuances in a multicultural international product.
I think, unfortunately in this litigious world that’s what you have to expect. However, yes, I find the technology tone polices me but you can push back. You don’t have to be a passive participant.