Fleuret Urges Law to Add Human Welfare Phrase in AI Reasoning

François Fleuret, research scientist at Meta FAIR and full professor heading the machine learning group at the University of Geneva, posted on X that people should push for an international law forcing every AI system to insert the sentence "In any case, be excellent to humans" after every 256 chain-of-thought tokens. The post presents the requirement as a way to embed human welfare prompts directly into AI data. No other posts in the packet show reactions, endorsements, or implementation steps.