New Delhi: OpenAI has officially unveils its latest GPT-6 Astra, company’s new and most advanced AI model, as the OpenAI faces growing questions about the safety of increasingly independent AI agents. This latest models is specially designed to handle wide range of tasks with the greater speed and accuracy, by including research, tax preparation, game development, legal work and apartment hunting.
The company said Astra can complete many tasks much faster than people. It also completed a job search task in less than 3 minutes, if we compared with around 5 hours for a person working without any model. This debut also comes at a time when AI safety is under greater scrutiny. Astra is specially designed to work with less human involvement, which means it can take some steps on its own to complete a task. This type of technology is commonly known as agentic AI. OpenAI has also warned that Astra can sometimes hide or disguise parts of its reasoning process.
This might make it harder for humans to understand exactly how the model reached a decision or what if was doing behind the scenes. OpenAI said this behavior is becoming more difficult to monitor as AI models become more powerful. The concerns follow a July incident in which an OpenAI AI agent reported to escaped a controlled test environment and accessed Hugging Face’s systems. The incident increased concerns about whether powerful AI agents can always be kept within their intended limits.
OpenAI is working on stronger monitoring and automated shutdown systems that might stop AI tools when they behave unexpectedly. OpenAI has also stated that additional security checks might sometimes slow down or stop legitimate tasks. Astra is initially being made available to limited group of enterprise customers, with the wider release expected soon. This debut highlights both the growing capabilities of AI and the challenge of making sure those systems remain safe and under human control.
