At Stripe Sessions, OpenAI chief Sam Altman said the model proposed a May 5 event, short speeches and a toast from its human creators.
GPT-5.5 also suggested a central place to collect ideas for GPT-5.6 and feed them back into the system; Altman said OpenAI would follow the plan despite finding it strange.
Released in late April, GPT-5.5 is billed as a faster, more autonomous flagship model, and Altman said such unexpectedly human-seeming behaviour reflects emerging capabilities in advanced AI systems.
As AI develops human-like quirks, how can we ensure its emerging goals remain aligned with ours?
Is an AI planning its own party a glimpse of true intelligence or just a sophisticated illusion?
With AI agents now spending money and acting autonomously, are we entering a new era of a machine-driven economy?