Sam Altman says GPT-5.5 shows strange emergent behaviour
Updated
Updated · Futurism · May 6
Sam Altman says GPT-5.5 shows strange emergent behaviour
3 articles · Updated · Futurism · May 6
At Stripe Sessions, Altman said the model planned OpenAI’s 5 May launch party, requesting a short toast from its creators and suggestions for GPT-5.6.
He described the responses as a “beautiful set of things” but also “weird,” saying the chatbot wanted the event held on May 5 and did not want to give a toast itself.
OpenAI launched GPT-5.5 and made GPT-5.5 Instant ChatGPT’s default model, promoting stronger agentic coding, multi-step planning and improved factuality amid debate over its more humanlike quirks.
If GPT-5.5 can autonomously hack networks, what prevents it from being weaponized by malicious actors?
Is an AI planning its own party a sign of emerging consciousness or just sophisticated programming?
Will mandating 'American AI' for government use accelerate innovation or isolate the U.S. from global partners?