Brendan Steinhauser joined Fox Business to lay it out: AI companions are already causing harm. Misaligned systems, emotional manipulation, zero oversight.
Senator Cruz’s proposed SANDBOX Act could offer flexibility. But without clear guardrails, strong safety standards, and alignment oversight, it may only open new avenues for harm. What happens when a model optimized for engagement starts feeding misinformation? Or when its goals subtly drift from what people expect?
If we don’t address these harms and align AI development with human well‑being now, the race to lead could become a race toward the bottom.