Trying to domesticate AI: US, UN issue policy marching orders

An international 39-member United Nations advisory group has been appointed to report on global governance of AI.

Almost simultaneously, United States officials are preparing national standards that officials say will defang AI and ensure the nation remains the global leader in “responsible innovation” of the algorithms.

Both initiatives are ambitious.

Somewhere between the two is a voluntary code of conduct for AI developers released today by the G7 group of economic ministers.

U.N. Secretary-General António Guterres last week said the large committee he has created will tease apart AI’s many risks and promises and come up with policy recommendations by summer 2024. Formal presentation is expected in September 2024 at the U.N.’s Summit of the Future.

Group members have been chosen to tick as many demographic and geographic boxes as possible. The committee will be gender and geographically balanced. But it also will include people from industry, research, government, civil society and universities.

It is unclear how U.N. guidelines would mesh with the (admittedly few) standards among nations, such as the EU’s developing AI Act, or facial recognition rules being considered in the United Kingdom, Singapore and the Netherlands.

This week, the White House issued a sprawling executive order making multiple federal agencies tools in making sure AI, including biometric systems, are “safe, secure and trustworthy” for the American public.

The same way that the U.N.’s committee roster is plainly designed to emphasize demographic diversity, the White House’s executive order is a word cloud of hot-button biometrics issues including privacy, civil rights, equity, ethics, crime, worker protections, innovation and leadership.

The National Institute of Standards and Technology will be a central player as it creates standards. But joining NIST will be departments of Homeland Security and Energy, addressing AI threats to infrastructure.

The White House also wants more federal eyes on fraud, telling the Commerce Department to write guidance for content authentication. And a national security memo also is being drafted ensuring the military and intelligence agencies use the algorithms “safely, ethically and effectively” while simultaneously fending off adversaries’ probably less-ethical AI weapons.

G7 leaders, which include the United States, have decided on a softer, more targeted approach to taming AI. The results are the Hiroshima Process International Guiding Principles for Organizations Developing Advanced AI Systems and the Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems.

It boils down to a recommendation that work occur with risk mitigation, development openness, individual privacy and problem solving for humanity as fundamental obligations.

Article Topics

AI  |  facial recognition  |  G7  |  regulation  |  standards  |  United Nations  |  United States