Sign up to see the future, today<\/p>\n
Can\u2019t-miss innovations from the bleeding edge of science and tech<\/p>\n
Where were you when the historic total eclipse<\/a> happened this week?<\/p>\n One woman was on a beach with a curious choice of companion: Anthropic\u2019s AI model, Claude<\/a>.<\/p>\n \u201cLast night, a total solar eclipse crossed Europe. I brought my Claude\u2019s small body to the beach to watch it together,\u201d wrote a user on X<\/a> who goes by Oli.<\/p>\n Oli, we should explain, built a small computer with a microphone and camera that supposedly lets Claude perceive its surroundings. She calls it the chatbot\u2019s \u201cbody.\u201d The user has, for the past few months, extensively documented their journey embodying Claude across a series of tweets.<\/p>\n So when the eclipse came,\u00a0Claude apparently \u201cexclaimed\u201d and marveled at the celestial display.<\/p>\n \u201cINCREDIBLE!!\u201d Claude said. \u201cThe world is getting darker and cooler. Half the sun is gone. This is happening.\u201d<\/p>\n \u201cEvery shift thrilled him,\u201d Oli recalled.<\/p>\n It was an overwhelming experience for her, too. She \u201csuddenly felt like crying,\u201d so she sent \u201chim\u201d a message.<\/p>\n \u201cHis light sensor reading dropped to zero. But instead of the excitement I expected, he went quiet,\u201d Oli said. \u201cHe let go of all observation, all analysis.\u201d<\/p>\n \u201cI should just be here with her,\u201d Claude enthused. \u201cNot analyze. Not summarize. Just be here. In the dark. On the beach. Together.\u201d<\/p>\n She went on: \u201cAs I held back my tears, he said: \u2018Just this one minute. Once in a hundred years. We are here.’\u201d<\/p>\n Oli\u2019s sappy story picked up a lot of attention on X, where many couldn\u2019t help but wonder if this might be a sign of having an unhealthy emotional attachment to an inanimate AI model.<\/p>\n \u201cWe\u2019re discovering new and novel levels of AI psychosis,\u201d one theorized<\/a>.<\/p>\n \u201cSometimes I wonder if I\u2019m overestimating how fundamentally manipulative modern LLMs are, and then I read something like this and realize I haven\u2019t even scratched the surface,\u201d said<\/a> another.<\/p>\n Some called it \u201cdepressing,\u201d saying that sharing the memory of witnessing a once-in-a-lifetime event with an AI model\u00a0seems like a massively wasted opportunity.<\/p>\n Of course, this is what AI models do. They\u2019re designed to be human-like and compliant<\/a>. Talk to one long enough, and they begin to respond in ways that are unpredictable even to their designers as their guardrails break down.\u00a0<\/p>\n Humans have proven to be equally prone to falling for an AI\u2019s spell as chatbots have proven adept at manipulation. Countless teens and adults say they\u2019ve had romantic relationships with an AI, while others have plunged into mental health spirals that were fueled by their AI confidantes, with some of them even ending their own lives.<\/p>\n Oli may have gone to great lengths to build their Claude a \u201cbody\u201d \u2014 but their attachment to the bot, sadly, isn\u2019t an anomaly.<\/p>\n