White House Demands Companies Share AI Security Risks with Feds

(IntegrityPress.org) – President Joe Biden has issued a new executive order concerning artificial intelligence, which is the US government’s first foray into the AI domain. The order mandates new assessments for AI safety, guidance on equity and civil rights, and looks into the impact on the labor market because of AI’s implementation. The order also maintains that companies must share all this information with the federal government.

President Biden revealed during a White House speech on Monday, October 30 that his meeting with Senate Majority Leader Chuck Schumer and a bipartisan committee on Tuesday aims to emphasize the necessity for congressional action.

The senior administration official noted that 15 major US technology companies have agreed to adhere to voluntary AI safety commitments but emphasized that this is insufficient. Monday’s executive order takes a significant step towards introducing concrete regulations for the development of this technology, especially regarding companies.

The executive order also further develops commitments that were previously obtained from leading AI companies by the White House, marking the first substantial binding government action concerning AI technology. It coincides with the AI safety summit on November 2 which is hosted by the UK.

The executive order requires major corporations to share safety test results with the US government before officially launching AI systems. The National Institute of standards and Technology is prioritized regarding techniques in developing standards for “red-teaming” AI, which involves stress-testing system defenses and identifying potential issues. The Department of Commerce will establish standards for watermarking AI-generated content as well.

The order also addresses training data for extensive AI systems and highlights the need to assess how agencies collect and use commercially available data, especially when it contains personal identifiers, including data obtained from data brokers.

The senior administration official clarified that the most ambitious timeframes for safety and security aspects of the order involve a turnaround time of about 3 months, while other aspects may require close to a year. This executive order follows various measures taken by the White House in recent months, aiming to provide platforms for discussions on the pace of AI development and the formulation of suggested guidelines.

Copyright 2023, IntegrityPress.org