OpenAI has told dozens of government agencies, universities and other organizations that their websites may have been impacted by improper visits from its AI models during evaluations, the company disclosed in a blog post published Friday.