The Department of Energy (DoE) is working on finalizing an Artificial Intelligence Risk Management Playbook, and an agency official said on March 9 that it should be released sometime in fiscal year (FY) 2023.
Dr. Pamela Isom, the Director of DoE’s Artificial Intelligence and Technology Office (AITO), said that her office has been working on the playbook with the National AI Initiatives Office, the National Institute of Standards and Technology (NIST), the Department of Defense (DoD), NASA, and others.
“The AI Risk Management Playbook isn’t ready for public consumption yet,” Isom said at a FCW webinar. “It is due to release in FY2023. So, stay tuned on that. Right now, it’s for DoE internal use only. However, we are working with NIST, so NIST will be integrating aspects of the playbook and what they’re doing with the framework.”
“But if you just hang in there, we’re going to make the playbook available in FY2023, which is not too far away,” Isom said.
Isom said it’s important to have a robust AI Risk Management Framework due to the fact that AI is being used in critical infrastructure and is, as a result, vulnerable.
“We use AI to help to defend our critical infrastructure, so our infrastructure can be vulnerable,” Isom said. “AI is particularly vulnerable because of the vast amounts of data, [and] the models – we call it adversarial attacks or adversarial AI.”
“The models are vulnerable,” she continued. “They can be distorted from the original intent. … And we may not even know that because of the fact that the models have been deployed. So, a part of our responsible and trustworthy principles is that we stay close to the models after they are deployed.”
