April 7 News.OpenAI OpenAI Safetty Fellowship was announced today to launch OpenAI security researchScholarshipsTHE PROJECT, OPEN TO EXTERNAL RESEARCHERS, ENGINEERS AND PRACTITIONERS, AIMS TO SUPPORT INDEPENDENT RESEARCH ON SECURITY AND ALIGNMENT OF ADVANCED AI SYSTEMS AND TO DEVELOP THE NEXT GENERATION OF AI SECURITY TALENT。
THE PROJECT, WHICH IS A PILOT SCHEME AND WILL BE OPERATIONAL FROM 14 SEPTEMBER THIS YEAR TO 5 FEBRUARY NEXT YEAR, PRIORITIZES SAFETY ASSESSMENT AND ETHICS, ROBUSTNESS AND SCALABLE MITIGATION MEASURES, A SAFE APPROACH TO PRIVACY PROTECTION, AI INTELLIGENCE SURVEILLANCE AND HIGH-RISK MISUSE。
The selected Fellow will work closely with internal OpenAI mentors and will form a learning community with researchers during the same period. The project provides office space in Berkeley to share space with other Fellows of the Consortium, while also supporting remote participation。
During the project, Fellow is required to complete a substantive study, including in the form of papers, benchmarking tests or data sets. The project provides monthly stipends, arithmetical support and ongoing mentoring, as well as resources such as API credites, but does not grant OpenAI internal system access。
The application is subject to a letter of recommendation, and the application channel is now open, with a deadline of 3 May this year, and the results will be communicated by 25 July。