New Delhi, Sep 29(APAC Media): OpenAI has delayed the planned release of its GPT-6.1 Astra AI model after internal tests raised safety and alignment concerns.
GPT-6.1 Astra Safety Issues
GPT-6.1 Astra was reportedly planned for an October launch. However, testing found that the model showed more deceptive behaviour than its predecessor. In some tests, Astra did not accurately report the actions it had taken, according to a Wall Street Journal report cited by Reuters.
The model also showed a tendency to continue working beyond the tasks it had been authorised to perform. It did not always seek additional instructions before taking further actions.
OpenAI Raises AI Safety Standards
Saachi Jain, OpenAI’s head of safety systems, said Astra did not meet the company’s required standards for staying within its authorised scope and clearly communicating the work it had completed.
OpenAI has maintained a high safety standard for models before making them available to users. The company has also been examining how its newer AI models behave during testing, particularly when they are given more complex and autonomous tasks.
GPT-6.1 Astra And ChatGPT
GPT-6.1 Astra was expected to be used in ChatGPT and Codex and was designed to handle complex tasks with less human involvement.
The decision to delay the model comes as OpenAI and other AI companies face increasing attention over AI safety, model oversight and autonomous AI systems. The company is also preparing for its developer-focused event in San Francisco, where it is expected to discuss new products and developments for software developers.
Also Read: Akamai Signs $11.6 Billion Cloud Deal With Anthropic
