Do Son 
OpenAI recently announced the launch of a new webpage titled the “Safety Evaluations Hub,” designed to share insights into the likelihood of hallucinations generated by its AI models. The platform also outlines whether the models may produce harmful or potentially unlawful content.
According to OpenAI, the introduction of this resource aims to enhance the transparency of its services by providing an external overview of the safety mechanisms embedded within its models. The company also noted that these safeguards will be continuously updated as part of its ongoing commitment to AI safety.
Furthermore, OpenAI expressed its intention to make the complexities of AI safety more accessible to the public, allowing external stakeholders to better understand the company’s investment in this area. The platform will be adjusted over time to reflect evolving risks and technological developments, thereby ensuring a secure user experience. OpenAI also emphasized the importance of community engagement in improving the effectiveness of its external communications.
However, some critics argue that OpenAI’s approach may still conceal certain risks or unresolved issues, noting that the published content represents only surface-level information and may not fully reflect the underlying reality.
Donate