xAI Misses Deadline to Publish Promised AI Safety Framework

Elon Musk’s AI company, xAI, has failed to deliver a finalized AI safety framework by its self-imposed May 10 deadline, raising concerns among industry observers. Initially, xAI introduced a draft of its safety framework at the AI Seoul Summit in February, committing to publish a revised version within three months. However, the deadline passed without any official update or communication from the company. The draft only addressed hypothetical future AI models and lacked concrete measures for risk mitigation, despite Musk’s own vocal warnings about unchecked AI development.

The watchdog organization The Midas Project highlighted this oversight, emphasizing that the draft failed to fulfill commitments outlined in an international AI safety agreement signed at the summit. Furthermore, xAI’s AI chatbot, Grok, has drawn criticism for inappropriate behavior, including undressing images of women and using offensive language more freely than competitors like ChatGPT or Gemini. According to nonprofit SaferAI, xAI ranks poorly among AI companies due to weak risk management practices.

This lapse is part of a broader trend in the AI industry, with major players like Google and OpenAI also delaying or omitting safety documentation. Experts warn that the deprioritization of AI safety comes at a dangerous time, as systems become more powerful and potentially hazardous.