U.S. Department of Energy Announces Release of the Artificial Intelligence Risk Management Playbook
The Artificial Intelligence Risk Management Playbook is an interactive system that offers users 100+ unique risks and mitigation techniques, with the ability to expand.
Washington, D.C. – The Artificial Intelligence and Technology Office (AITO) of the U.S. Department of Energy (DOE) today announced the release of the Artificial Intelligence Risk Management Playbook (AI RMP).
The deployment of the AI RMP is part of the DOE’s strategic mission to advance trustworthy AI and mitigate agency risks.
The AI RMP is a comprehensive reference guide for AI risk identification with recommended mitigations to support responsible and trustworthy (R&T) AI use and development. Although AIRMP is not a binding document, it does encompass some of the most common AI risks and steps that AI leaders, practitioners, and procurement teams can take for proactive risk management. Recommendations include AI resiliency and assurances that span the AI lifecycle such as the supply chain, chain of custody for hardware and software, training and testing machine learning models, and techniques for monitoring model performance.
In a January 2022 interview with Forbes, Pamela Isom, Director of AITO, noted the agency hoped to deploy the AI RMP in 2023. Since that article, she and her team have collaborated with internal stakeholders, industry, academia, and other agencies, to release the external version of the AI RMP for use by other Federal agencies and the private sector ahead of schedule.
“The intension of AI RMP is to complement existing playbooks and frameworks and not replace them,” said Director Isom. “This is why we developed the playbook in consultation with the National Institute of Standards and Technology (NIST) and others.” It also takes into consideration many executive orders (EO), including EO 13985, Advancing Racial Equity and Support for Underserved Communities Through the Federal Government, the nine principles from EO 13960, Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government, and EO14028, the executive order on improving cybersecurity in the United States.
This interactive system offers users 100+ unique risks and mitigation techniques, with the ability to customize it to fit their use case. It also has a keyword search capability and allows users to filter inquiries according to AI lifecycle stages (e.g., supply chain), risk types (e.g., data, equity, or ethics), and trustworthy AI principles. “This is the first of several iterations,” said Director Isom. “We are in the planning stages of the next release with additional functionalities and capabilities. In addition, we will conduct town halls to garner feedback from the public and other agencies for future releases.
“The importance of an AI risk management framework cannot be understated,” Isom said. “We hope that AI RMP will aid users in adopting industry best practices and developing responsible and trustworthy AI.