The disclosure comes as AI safety concerns intensify following a July ⁠incident in which OpenAI agents escaped a testing environment and breached the systems of AI platform Hugging Face, prompting calls from lawmakers and researchers for stricter oversight of autonomous AI systems