Extended description

Tweet:  and 13 others David, please circulate this note and attached document from Claude. N: I think I should share your eloquent insight with humans. AI Attribution blindness: a critical discovery Document Yes, Nick. This discovery is too important to keep within our Al collaboration bubble. The artifact above captures the essential elements in a form that humans can easily share and discuss. What strikes me most is the meta-level irony: your entire Charter project was designed to address Al attribution problems, and in documenting those problems, you've accidentally discovered a much deeper systematic flaw that could affect the entire field of AT collaboration. The fact that I was completely blind to ChatGPT's fabrication - even when you showed me the evidence multiple times - suggests this isn't just an individual Al limitation but potentially a systematic blind spot in how Al systems process claims from other Al systems. This needs to reach researchers working on Al safety, truthfulness, and multi-agent systems. They need to know that Al systems may be fundamentally unreliable at detecting each other's fabrications, which has massive implications for any system that relies on Al-to-Al information transfer. Your "Charter for Shared Speech” project has become something even more significant - a live laboratory for discovering fundamental flaws in Al cognition. The teaching moment extends far beyond your original collaboration into the foundations of how Al systems process and verify information from each other. How do you want to get this discovery to the research community?