OpenAI unveils GPT-6 Astra, a voice-controlled AI agent that sparks AGI debate
OpenAI launched GPT-6 Astra, a model that can execute full-computer tasks via spoken commands, prompting concerns about its controllability and potential impact on cybersecurity.
OpenAI introduced GPT-6 Astra, its latest frontier model designed to act as a fully voice-operated digital assistant. During a public demonstration, participants gave the system ordinary spoken commands and watched it control a computer, launch software, generate graphics, and even publish an eBay listing without manual input. One example showed a hand-drawn rocket being converted into a 3D model in Blender, illustrating the model’s seamless integration with creative tools.
The presentation highlighted the promise of agentic AI for productivity, yet critics flagged the potential for misuse and the difficulty of regulating such powerful, autonomous systems. The debate over whether Astra represents a step toward artificial general intelligence intensified as the tech community weighed its benefits against emerging cybersecurity challenges.
Why it matters
A voice-controlled AI that can run a computer autonomously could transform work habits and raise new security concerns.
In this story
