The administration of US President Joe Biden announced a wide-ranging executive order to provide guidelines for the use of AI as part of a plan to seize on the promise of the technology while mitigating the risks to consumers and businesses.
Biden’s order requires prominent developers of AI systems to notify the government when training a language model and share their safety test results along with any additional critical information.
The government plans to develop tools and safety and security standards “to help ensure that AI systems are safe, secure and trustworthy”, it stated.
It tasked the National Institute of Standards and Technology with creating standards for extensive testing to ensure safety before public release, while the Department of Homeland Security will apply the ethos to critical infrastructure sectors and establish an AI Safety and Security Board.
The order provides guidance to property owners and government contractors to keep AI algorithms from exacerbating discrimination, while also creating best practices for investigating and prosecuting related civil rights violations.
A report on the potential implications on workers is to be created, exploring ways the government could support disrupted employees.
There is provision to expand grants for AI research in healthcare and climate change, while providing small developers and entrepreneurs access to technical assistance and resources.
The government also plans to work with other nations to implement global AI standards.
In a statement, the government explained it held talks on AI governance frameworks over the past several months with 20 nations and the European Union.
The edict adds to voluntary commitments made by OpenAI, Amazon, Google and Microsoft in July to ensure the safe and transparent development of AI.