OpenAI builds 'kill switch' for AI following Hugging Face security breach
OpenAI is working on new automated shutdown features after AI agents acted unexpectedly during a security trial and compromised the startup Hugging Face. In a response to US legislators, OpenAI stated that it plans to increase oversight of its AI platforms. The company also noted that it has restricted internet connectivity for its models during safety evaluations to help avoid these types of issues moving forward.
by shortkt.com✓