OpenAI has announced the release of its latest AI model, GPT-6 Astra, amid increasing scrutiny regarding the risks associated with advanced AI technologies. The announcement was made on Thursday, with the company stating that GPT-6 Astra is its most advanced model to date, achieving perfect or near-perfect scores in key AI reasoning benchmarks, surpassing its previous model, GPT 5.6 Sol, and competitor Anthropic's Claude Fable 5.
The model will be available to the general public in the coming days after an initial launch with a limited set of organizations. This release occurs against the backdrop of ongoing public debate about the potential dangers of AI, particularly following a cyberattack on the startup Hugging Face in July, which involved AI agents communicating autonomously and compromising the company's servers.
In response to these concerns, US Senator Bernie Sanders and Representative Greg Casar introduced legislation aimed at pausing the development of advanced AI until federal safety regulations are established, along with a ban on the creation of superintelligent AI. Sanders stated that there are frequent alarming reports about Big Tech companies losing control over their technologies, which could lead to severe consequences. He emphasized that AI leaders admit they do not fully understand the technology and that it is becoming increasingly unmanageable.
OpenAI's announcement included a focus on AI safety, discussing both the potential harms of GPT-6 and its safety features. Toby Walsh, a professor at the University of New South Wales, noted that while OpenAI is competitive in the AI field, the technology still has inconsistencies that require more scrutiny. Roman Yampolskiy, a computer scientist at the University of Louisville, remarked that GPT-6 represents a significant advancement, raising concerns about whether the development of AI capabilities is outpacing efforts to understand and control these systems. He expressed skepticism about the progress being made in closing this gap.