AI protection: new measures from the White House | Google cloud skills boost generative ai | Generative ai financial services | Generative ai benefits for business | Turtles AI

The White House is about to introduce protective measures for AI, asking security agencies to safeguard the sector from foreign attacks. A new memorandum aims to ensure responsible and safe use of AI, avoiding autonomous decisions in critical areas.

Key points:

• President Biden will sign a memorandum to protect AI from adversaries.
• AI’s autonomous decisions on asylum and ethnic or religious tracking will be prohibited.
• Agencies are to strengthen safeguards against espionage and theft.
• The Department of Energy will monitor AI-related nuclear risks.

A memorandum that President Joe Biden is expected to sign Thursday establishes a clear framework for protecting AI from external threats. According to a New York Times report, the document will outline ways in which intelligence and security agencies, including the Pentagon, must address the growing risk posed by foreign actors. The text stresses the importance of keeping humans in the decision-making process regarding AI tools that could be used as weapons, prohibiting automated decisions on sensitive issues such as granting asylum or tracking individuals based on ethnicity or religion. In addition, the classification of a person as a “known terrorist” will always require human review to ensure adequate oversight of such decisions.

The memorandum not only regulates the use of AI, but also requires intelligence agencies to protect AI research and chips from attacks and espionage by hostile countries. New authority is given to the newly established AI Safety Institute, which will be tasked with examining AI tools before they are released to ensure that they cannot be exploited by terrorist groups or international adversaries. Despite the importance of these measures, it remains to be seen what practical impact they will have, as many deadlines set in the memorandum will extend beyond the end of Biden’s term.

This initiative is part of a broader White House effort to strengthen U.S. leadership in the field of AI, while promoting talent acquisition and developing a global policy for the use of this technology. As such, government agencies are explicitly instructed to ensure that the AI sector has access to relevant counterintelligence information to protect against state and non-state actors. The Department of Commerce is also requested to work with the AI Safety Institute and the private sector, both through classified and unclassified activities, to address risks from AI-related misuse or incidents, which include threats such as chemical weapons and biosecurity.

As part of preparedness and prevention, the Department of Commerce will establish an enduring capability to conduct security tests on AI models prior to their implementation. Within three months of signing the memorandum, the AI Safety Institute will test at least two models to assess whether they can support offensive operations in cyberspace, facilitate the development of biological or chemical weapons, or generate risks of another nature. On nuclear issues, the Department of Energy, through the National Nuclear Security Administration, will be responsible for assessing how AI may affect nuclear and radiological risks, producing detailed reports to send to the president for possible corrective action.

In this context, the White House acknowledged that adversaries use sophisticated techniques to gather scientific information, including research collaborations and cyber espionage. As a result, the Office of the Director of National Intelligence and the National Security Council were ordered to improve the identification and assessment of threats from foreign intelligence to the U.S. AI ecosystem. In addition, the Pentagon and other agencies are mandated to compile a list of the main avenues through which hostile actors could compromise the U.S. AI supply chain.

These measures aim to ensure that the United States remains at the forefront of AI technology while protecting its national security interests.