OpenAI stated in April 2025 it may "adjust" safety requirements if a competing lab releases high-risk AI without similar protections. Reports indicate OpenAI reduced safety testing time from months to days before releases.