OpenAI was planning to open this new model to everyone’s access. This means that the model is shared not only through an API, but can be downloaded directly to computers. However, Altman stressed that this approach has great risks:
“Once the weights are published, there is no return. Therefore, more security assessments must be made.”
According to Altman, the potential of abuse of artificial intelligence – especially in open source versions – is still not sufficiently controlled. Therefore, OpenAI will tighten security controls to prevent the model from being used in scenarios that may be harmful to society.
Balance between security and transparency
In the world of technology, the decision of OpenAI to postpone an important discussion also brought up:
On the one hand, companies like Meta encourage the community contribution by developing artificial intelligence as open source.
On the other hand, companies like OpenAI are trying to prevent abuse by controlling powerful models as centrally.
Although Altman appreciates the potential of open -source development, he argues that the opening of powerful models to unlimited access can lead to serious risks. Especially in areas such as disinformation, malware production or automatic attack systems, malicious people can abuse these models.
New Date Not Given
OpenAI did not explain a definite date on when the postponed model would be released. However, the company clearly stated that it will not share any sharing before all security checks are completed.
