National Security and AI: Insights from Deputy Advisor Anne Neuberger

Government Deputy National Security Advisor Anne Neuberger on addressing the security threats of AI

The increasing prominence of artificial intelligence (AI) has raised concerns among policymakers about its potential security risks. In a recent interview, Anne Neuberger, the deputy national security advisor for cyber and emerging technologies in the Biden administration, discussed the White House’s efforts to address the security implications of AI.

The Dual Nature of AI

Neuberger acknowledged that AI, particularly large language models, presents both risks and benefits. While these technologies have the potential to revolutionize various sectors such as drug discovery, sustainable infrastructure, design, music, and art, they also introduce security concerns and risks. The key is to carefully manage these risks while harnessing the benefits.

Assessing the Risks

The discussion then turned to the existential risks associated with AI. Neuberger emphasized the need to understand the most significant risks and develop thoughtful approaches to address them. The Biden administration recently hosted a group of AI companies at the White House to discuss voluntary controls that could mitigate risks, enhance transparency, and address core elements of these risks. However, Neuberger noted that there is still much to learn about these technologies and called for close partnership with the companies building them to ensure comprehensive risk management.

Explaining AI to the Public

When asked about how she explains AI to people outside the field, Neuberger shared an example from her visit to Cincinnati and Chicago. She described how AI is being used to improve government services, specifically in a 911 center, where AI is employed to train operators by analyzing historical data to determine effective responses to different types of emergency calls. Neuberger’s personal experience in cybersecurity has shaped her understanding of technology’s dual nature, and she emphasized the importance of understanding the significant risks while also leveraging the benefits it offers.

Addressing Current Security Concerns

In terms of the current security concerns related to AI, Neuberger highlighted several areas of focus. One primary concern is the potential use of AI in designing malware and deep fakes, as well as customizing spear-phishing attacks and bio-weapons. These capabilities could lead to more effective and rapid cyberattacks, precision in disinformation campaigns, and the potential for manufactured crises. The recent incident involving the dissemination of a fake image of an explosion at the Pentagon serves as a reminder of the need to question the authenticity of information and rapidly respond to disinformation.

Watermarking and Trust in AI-Generated Content

Neuberger discussed the concept of watermarking as a potential solution to address the issue of AI-generated content. She highlighted the importance of private sector leadership in implementing watermarking technologies that can authenticate AI-generated content. The private sector AI firms possess the necessary expertise to distinguish between generated and real content. Neuberger also compared watermarking to how web browsers display green padlocks to indicate secure connections, suggesting that similar indicators could be developed to verify the authenticity of AI-generated content.

Arms Control Frameworks and Regulating AI

Regarding the regulation of AI, Neuberger recognized the importance of international cooperation and highlighted ongoing policy discussions to address AI-related threats. She mentioned the need for robust testing and security measures through partnerships with AI companies to counter threats and safeguard against harms in the short term. While longer-term efforts are being considered, focusing on the immediate threats should remain a priority.

Cybersecurity Partnerships and Lessons from Ukraine

Neuberger emphasized the power of partnerships and shared the lessons learned from the United States’ collaboration with Ukraine in addressing cyber threats. Ukraine’s experience with Russia’s offensive cyber program led to significant improvements in cybersecurity through international support and the involvement of the private sector. Neuberger also highlighted the administration’s efforts in international cyber partnerships, such as the International Counter Ransomware initiative, focusing on countering cybercrime and promoting cybersecurity across critical infrastructure.

Chinese Cyber Threats and Relations

Regarding Chinese cyber threats, Neuberger differentiated between espionage for research and technology theft and state-sponsored cyber actors potentially targeting critical infrastructure for disruptive operations. While she refrained from providing specific interpretations of recent incidents, she acknowledged that China’s cyber capabilities pose serious threats to both the United States and global critical infrastructure. Neuberger discussed the administration’s proactive approach to improving the security of critical infrastructure and the ongoing tension in U.S.-China relations.

Collaboration and Coordination in Emerging Technology and Cybersecurity

Neuberger emphasized the importance of collaboration between traditional national security agencies and agencies familiar with specific sectors, such as energy, water systems, and transportation. The coordination between these entities allows for a comprehensive approach to address cybersecurity threats, including those related to critical infrastructure. She highlighted the privilege of her role in coordinating policy across the U.S. government and expressed appreciation for the collaboration between different agencies.

Conclusion

Neuberger’s insights provide valuable perspectives on the challenges associated with securing AI systems and effectively mitigating the risks they pose. She emphasizes the importance of understanding both the promises and risks of AI and the need for collaboration between the public and private sectors to address these issues. As AI continues to advance and permeate various aspects of society, policymakers, industry leaders, and individuals must actively engage in discussions surrounding its security and devise appropriate measures to ensure its responsible and secure deployment.