OpenAI Prioritises Safety Aspects in AI Model Development

Jakarta (ANTARA) - OpenAI, the artificial intelligence research and development company, is prioritising safety aspects in the development of its AI models amidst increasing concerns regarding the potential risks posed by the technology.

A number of global technology leaders have expressed the need to slow down AI development, as the rising risks are deemed alarming.

Speaking at a discussion in Jakarta on Wednesday, OpenAI’s Head of Policy for Southeast Asia, Sandy Kunvancatagarn, stated that OpenAI is employing a multi-layered security approach in developing AI models and is continuously refining this as AI capabilities evolve.

“Safety aspects are built into every stage, starting from how we train the AI models, the post-training phase, to developing the various tools and products built upon those models,” he said.

OpenAI also provides reports and discloses information to assist researchers and other institutions in understanding the capabilities and limitations of the developed AI models.

Furthermore, the company utilises public feedback as a support mechanism in the AI evaluation and development process.

For instance, the company uses community input to refine safety features when developing products intended for children and adolescents.

Sandy noted that interacting with the community, including parent groups, provides insight into public concerns regarding AI usage.

“All of this is to say that the capabilities of these AI models are increasing very rapidly, and we are refining our security approach in real time,” he added.