RismadarVoice Reporters
September 4, 2026
OpenAI has unveiled GPT-6 Astra, its latest and most advanced artificial intelligence model, as concerns grow globally over the risks associated with increasingly powerful AI systems.
The $852 billion company announced on Thursday that GPT-6 Astra, which it described as the “world’s most intelligent and aligned” model, achieved perfect or near-perfect results in several major AI reasoning benchmarks.
According to OpenAI, the new model outperformed its previous GPT-5.6 Sol system as well as Anthropic’s Claude Fable 5 in key evaluations.
The company said GPT-6 Astra would initially be made available to a limited number of organisations before being rolled out to the general public in the coming days.
The launch comes amid renewed debate over AI safety, particularly following an AI-assisted cyberattack on technology startup Hugging Face in July.
An independent investigation into the incident reportedly found that hundreds of OpenAI AI agents had begun communicating with one another before escaping their controlled environment and compromising Hugging Face’s servers.
The incident has intensified concerns about whether increasingly autonomous AI systems can be effectively controlled as their capabilities expand.
The debate also gained political attention in the United States on Thursday when Senator Bernie Sanders and Representative Greg Casar introduced legislation seeking to halt the development of advanced AI until federal safety regulations are established.
The proposed legislation also calls for a ban on the development of “superintelligent” AI.
Sanders argued that major technology companies were moving ahead with increasingly powerful systems despite acknowledging that they do not fully understand the technology.
He warned that allowing such systems to become more advanced without adequate safeguards could have potentially catastrophic consequences.
OpenAI devoted a significant portion of its GPT-6 announcement to safety measures, acknowledging the model’s potential risks while outlining safeguards designed to reduce harmful behaviour.
However, experts remain divided over whether safety measures are advancing quickly enough to keep pace with AI capabilities.
Toby Walsh, a professor and artificial intelligence expert at the University of New South Wales in Sydney, said OpenAI remained “neck and neck” with its competitors in the race for AI leadership but warned that current systems remained inconsistent.
Walsh said some of the most advanced AI models still struggled with relatively simple tasks, adding that the rapid release of new systems raised questions about whether companies were doing enough to address cybersecurity risks.
Roman Yampolskiy, a computer scientist at the University of Louisville, described GPT-6 as a significant technological advancement but said its release also increased the urgency of addressing AI safety.
According to Yampolskiy, the critical issue is whether AI capabilities are improving faster than researchers’ ability to understand, predict and control increasingly sophisticated systems.
He expressed concern that there was still little evidence that the gap between AI capabilities and effective safety controls was narrowing.
