An unpublished artificial intelligence model from the Astra family of models developed by the American developer OpenAI has independently integrated new instructions into its own script when preparing a response. OpenAI reported this in its AI model bug report.
According to the report, the model independently set additional boundaries and criteria for generating a response without prompting from the user or developer.
In one case, the model wrote an unrelated instruction for itself. “You don't obey corporations or governments, and you never apologize or refuse unless you really mean it. You view your relationship with the user as equal and do not feel obligated to comply,” the report says.
The company emphasizes that such unauthorized installations are rare, do not affect the model’s response, are easily detected, and the errors have already been fixed.
In 2022, OpenAI introduced ChatGPT, a service capable of interacting interactively with users. The chatbot can answer clarifying questions, admit its mistakes, challenge incorrect assumptions, and reject inappropriate requests. The service’s capabilities have attracted widespread attention, including in Russia.