US officials optimistic on AI but warn of risks, abuse

SAN FRANCISCO -- Federal government officials at RSA Conference 2024 touted the enormous benefits of artificial intelligence but also emphasized the need to protect against risks and potential abuse of the technology.

Artificial intelligence and specifically generative AI once again dominated the world's biggest cybersecurity conference, and throughout the week government leaders weighed in on the technology and what it means for both the public and private sectors. In his RSA Conference 2024 keynote on Monday, Secretary of State Antony Blinken unveiled the State Department's U.S. International Cyberspace and Digital Strategy, which outlines how the U.S. government plans to engage and partner with other nations on a range of technology issues, including AI.

"When it comes to AI, again, as confident as we are in its potential, we're deeply aware of its risks: from displacing jobs, to generating false information, to promoting bias and discrimination, to enabling the destabilizing use of autonomous weapons," he said during his keynote. "So we're working with our partners to prevent and address these issues."

Blinken highlighted President Joe Biden's executive order last fall to create standards for safe and secure development of AI, as well as the recent creation of the U.S. AI Safety Institute Consortium, which includes more than 200 private companies such as Google, Microsoft, Nvidia and OpenAI.

"The private sector is a critical partner in this effort, which is why we've worked with leading AI companies on a set of voluntary commitments, like pledging to security testing before releasing new products, developing tools to help users recognize AI-generated content," Blinken said.

The State Department also began piloting GenAI projects this year to assist with searching, summarizing, translating and even composing documents, which Blinken said frees up staff members to have more face time instead of screen time.

Alejandro Mayorkas, secretary of the Department of Homeland Security, also discussed applications for AI technology in DHS pilot projects. For example, one project combines all criminal investigation reports and uses AI "to identify connections that we would not otherwise be aware of," he said.

"What I would love is for this audience to take a look at DHS in five years and say, 'Wow, I cannot believe how they are using AI to advance their mission.' That is a redefining of the perception of government, not as slothful and labyrinthian but nimble, dynamic and really pushing the envelope ourselves," Mayorkas said.

However, there are significant risks from both internal and external use of AI, he said. To that end, DHS last month released safety and security guidelines for U.S. critical infrastructure organizations regarding AI usage, as well as potential outside threats. Those threats include AI-enhanced social engineering attacks such as deepfake audio and video.

But Mayorkas emphasized that organizations must also consider the risks associated with AI design and implementation. One thing that was made clear in the inaugural meeting of DHS' newly formed AI Safety and Security Advisory Board, he said, was that safe and responsible development of the technology go hand in hand. "We cannot consider the safe implementation to mean a potential perpetuation of implicit bias, for example," he said.