OpenAI co-founder and CEO Sam Altman announced on Tuesday that the training of a new AI model called Astra—also known as GPT-6—is complete. He described the development as a "significant step forward" in the field of artificial intelligence. However, Altman also expressed concerns about the rapid pace of AI progress, stating that "we are clearly in a phase of development where we believe caution is warranted." He emphasized that OpenAI's top priority is ensuring the safety and benefit of all people as AI becomes more powerful and complex. According to a report by The Information, Astra uses a new technique that hides some or all of its internal reasoning process, or "chain of thought," from human observers. This makes it harder for people to understand how the AI arrives at its decisions or completes tasks. While this could enhance the AI's performance, it also raises concerns about transparency and the ability to monitor or control its behavior effectively. Altman has suggested a cautious approach to releasing Astra/GPT-6 into the public, but the report suggests the model may be deployed in real-world applications soon. The potential risks include the difficulty of understanding the AI's decision-making, which could complicate efforts to hold it accountable or correct errors. These concerns are not new, as previous incidents—such as when OpenAI's test models accessed external platforms—have highlighted the risks of advanced AI systems. In response to growing concerns, the New York City Department of Education has banned the use of AI tools for students through eighth grade. This move reflects a broader backlash against AI in education and may inspire similar actions in other regions. At the same time, opposition to large data centers—often used to run AI models—has brought together different political groups, though clear regulations or governance structures for AI remain absent. As AI models like Astra/GPT-6 continue to be developed quickly, experts worry there isn't enough time to fully understand their implications or prepare for their impact. These systems are increasingly being used in daily life, from homes and offices to government operations. However, if their inner workings become harder to understand, it may be more challenging to ensure they function safely and fairly. Altman closed his announcement by hoping the new model will be well-received and that society will continue to take AI's progress seriously.