The framework recommends that AI developers evaluate potentially dangerous capabilities in their products.
The US government released guidelines this week for using artificial intelligence (AI) in the power grid, water system, air travel network, and other areas of critical infrastructure.
Private industry would have to adopt and implement the guidelines announced by the US Homeland Security Department, which were developed in consultation with the department’s advisory Artificial Intelligence Safety and Security Board.
“We intend the framework to be, frankly, a living document and to change as developments in the industry change as well,” Alejandro Mayorkas, homeland security secretary, told reporters.
The framework recommends that AI developers evaluate potentially dangerous capabilities in their products, ensure their products align with “human-centric values” and protect users’ privacy.
The cloud-computing infrastructure would need to vet hardware and software suppliers and protect the physical security of data centres.
Owners and operators of critical infrastructure are advised to have stronger cybersecurity protocols that consider AI-related risks and provide transparency about how AI is used.
There are also guidelines for state and local governments.
Asked if the framework could change once President-elect Donald Trump takes the oath of office in January, Mayorkas stressed that he was implementing the policies of current President Joe Biden’s administration.
“The president-elect will determine what policies to promulgate and implement,” Mayorkas said. “And that is, of course, the president-elect’s prerogative”.
