@claimedandunhinged: The 10% is real. The 2030 is real. They came from two different men, and the video glued them together. There is a video going around saying researchers give AI a ten percent chance of destroying humanity by 2030, backed by senior people at OpenAI and Anthropic. So we looked up who actually said what. The number came from the man who leads alignment science at Anthropic, and he said more than ten percent within a decade, which is a floor, not a ceiling. The date came from a different man, who quit Anthropic this week and walked away from his unvested shares to say it, and he put no number on anything. Then the heads of Anthropic and OpenAI said the labs must slow down, and Elon Musk agreed. What they promised was outside auditors. Not one of them said they would train less. Then we argue about China, competition, and whether you get to call the same machine autocomplete on Monday and a god on Friday. #AIboyfriend #AIcompanion

Sin & Sarah

Open In TikTok:

Region: CA

Monday 14 September 2026 23:01:10 GMT