In a new executive order issued by the US government, technological companies will now be obligated to share the results of their artificial intelligence (AI) system testing with the government prior to public release. This order was announced on Monday, […]
In a new executive order issued by the US government, technological companies will now be obligated to share the results of their artificial intelligence (AI) system testing with the government prior to public release. This order was announced on Monday, just before the important Artificial Intelligence Safety Summit taking place in the United Kingdom on November 1st and 2nd.
The directive states that as the administration works to advance this agenda domestically, it will also collaborate with allies and partners around the world to establish a strong international framework for managing the development and use of AI.
The White House directives regarding AI include the following:
- Companies developing AI models that pose a threat to national security, economic security, or public health and safety must share their safety testing results with the government.
- The government will prescribe guidelines for red team testing, where evaluators simulate adversarial actors during the testing process.
- Official guidance will be issued for labeling content created with the help of AI to prevent risks of harm caused by deepfakes and manipulations.
- New standards will be developed for testing biological synthesis to mitigate the threat of AI systems aiding the creation of biological weapons.
Jeff Zients, the Chief of Staff of the White House, stated that President Joe Biden has directed his staff to act swiftly on matters concerning artificial intelligence.
The White House spokesperson said, “We cannot move at the normal pace of government. We have to move quickly, if not faster than the technology itself.”
The White House has stated that sharing testing results for powerful AI models will “ensure that AI systems are safe, secure, and reliable before companies release them publicly.”
Under the provisions of the AI directive, the US Department of Commerce will issue guidelines for labeling and an AI-generated content framework to facilitate the distinction between authentic interactions and software-generated ones.
Regarding the labeling plans, the directive states, “Federal agencies will use these tools to make it easy for Americans to know that the communications they receive from their government are authentic – setting an example for the private sector and governments worldwide.”
The order also encompasses areas such as privacy, civil rights, consumer protection, and workers’ rights.
According to White House officials, the tasks outlined in the order will be implemented within 90 to 365 days, with security and safety being the top priorities.
A national security memorandum will be sent to the US military and intelligence community on how to use artificial intelligence in a secure and ethical manner. Congress will also be asked to pass legislation to protect Americans’ data privacy. Federal agencies will develop guidelines for assessing privacy-preserving techniques in AI systems.
The concern about bias is addressed in the order, providing guidance to landlords, federal social benefit programs, and federal contractors to prevent AI algorithms from exacerbating discrimination. The key immediate concern with AI systems is that they may unconsciously perpetuate underlying biases present in the data sets they are trained on. Best practices will be developed for the use of AI in the judicial system, including areas such as sentencing, predictive policing, and parole decisions.
The order also addresses the threat of job disruption by calling for the development of best practices to mitigate the harm of job loss, such as “preventing employers from underpaying workers, unilaterally evaluating job applications, or undermining workers’ ability to organize.” Government agencies will also receive guidance on the use of AI, including standards for protecting rights and safety.
The Federal Trade Commission, the competition watchdog, will be encouraged to use its authorities if any distortions are found in the AI market.
To support efforts in regulating AI worldwide, including discussions at the upcoming safety summit, the White House has announced that it will expedite the development of AI standards with international partners. Vice President Kamala Harris will represent the White House at the summit.