OpenAI has recently revealed its strategies to avert any severe consequences stemming from the advanced AI technology it develops.
The company has released a detailed 27-page “Preparedness Framework” document, which describes its efforts to monitor, assess, and mitigate “catastrophic risks” associated with state-of-the-art AI models.
These risks include the potential misuse of AI models in large-scale cybersecurity breaches or their application in developing biological, chemical, or nuclear weaponry.
Under this new preparedness framework, OpenAI maintains that the decision to launch new AI models initially lies with the company’s leadership. However, its board of directors has ultimate authority, including the power to overturn decisions made by the leadership team.
A professor from the Massachusetts Institute of Technology, Aleksander Madry, has temporarily left his post at MIT to lead OpenAI’s preparedness initiative. His role involves guiding a team of researchers responsible for identifying and meticulously tracking potential risks. They will be creating scorecards that evaluate these risks, classifying them into categories such as “low,” “medium,” “high,” or “critical.”
According to OpenAI, The preparedness framework is “only models with a post-mitigation score of ‘medium’ or below can be deployed,” and only models with a “post-mitigation score of ‘high’ or below can be developed further.”
The document is currently in a “beta” phase, as indicated by the company, with plans for regular updates based on feedback.
Earlier this year, a one-sentence open letter was signed by numerous leading AI scientists and researchers, including OpenAI’s Altman and Google DeepMind’s CEO Demis Hassabis. The letter emphasized that addressing the “risk of extinction from AI” should be a global priority, comparable to other critical risks such as pandemics and nuclear war.
