Frontier Model Forum Updates


Over the past year, the industry has driven significant advances in AI capabilities. As this progress has accelerated, new academic research into AI security is needed. To address this gap, the Forum and philanthropic partners are creating a new AI Security Fund to support independent researchers from around the world affiliated with academic institutions, research institutions and startups. Initial funding for the AI ​​Security Fund comes from Anthropic, Google, Microsoft, and OpenAI, and the generosity of our philanthropic partners, the Patrick J. McGovern Foundation, the David and Lucile Packard Foundation(^footnote-1), Eric Schmidt and Jaan Tallinn. Together, this amounts to more than $10 million in seed funding. We expect additional contributions from other partners.

Earlier this year, Forum members signed voluntary commitments on artificial intelligence at the White House, which included a promise to allow third parties to discover and report vulnerabilities in our artificial intelligence systems. The Forum sees the AI ​​Security Fund as an important part of meeting this commitment by providing funding to the external community to better assess and understand border systems. The global AI security debate and general AI knowledge base will benefit from a wider range of voices and perspectives.

The primary focus of the Fund will be to support the development of new model evaluations and techniques for red team artificial intelligence models to help develop and test techniques for evaluating the potentially dangerous capabilities of frontier systems. We believe that increased funding in this area will help raise security standards and provide insight into the mitigations and controls that industry, governments and civil society need to address the challenges posed by AI systems.

The fund will announce a tender in the next few months. The Meridian Institute will manage the Fund—their work will be supported by an advisory board made up of independent external experts, experts from AI companies, and individuals with grantmaking experience.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *