OpenAI is facing criticism for significantly reducing the time and resources allocated to safety testing of its powerful AI models, prioritizing speed to market over thorough risk assessment. Sources indicate that evaluations for new models like 'o3' now take days instead of months, raising concerns about potential risks. This accelerated release schedule is driven by intense competition in the rapidly expanding AI sector. While OpenAI claims to have improved efficiency through automation, critics argue that inadequate testing, particularly on final models, could lead to unforeseen and dangerous consequences. The lack of comprehensive global safety standards further exacerbates these worries.

Prepared by Jonathan Pierce and reviewed by editorial team.