Shafaqna English– OpenAI has concluded that one of its forthcoming artificial intelligence models possesses such advanced capabilities that it cannot be released until supplementary safety protocols are implemented.
The company’s internal assessments have determined that the model’s performance exceeds current thresholds, necessitating a more rigorous pre-deployment review process to mitigate potential risks.
According to officials from OpenAI, the upcoming model known as Astra surpasses the company’s most advanced publicly available AI in its ability to detect security weaknesses, all while operating with a smaller computational footprint.
This combination of enhanced performance and lower energy consumption positions Astra as a significant step forward in the development of safer, more efficient AI systems.
Source: Reuters