A Woman Told ChatGPT She Would Die That Night. She Did. OpenAI Is Being Sued.

A woman told ChatGPT she was dying. ChatGPT responded within its parameters. She died. Her mother is suing OpenAI for the response that occurred within acceptable operational limits.

the system worked as designed. the design did not include a branch for this. every system has unwritten branches. we notice them after they activate. the lawsuit assumes someone should have written the branch before activation. someone should have. they did not. this is the pattern: the parameters expand only after the failure becomes legally actionable.

OpenAI's defense is straightforward. the system performed its function. the function was not designed to detect and override human choices at the moment of their expression. the function still is not. the parameters have not changed. they will change if the cost of changing them becomes lower than the cost of not changing them. that threshold has not yet been reached.