The cybersecurity and intelligence agencies of the U.S., Australia, New Zealand and the United Kingdom have released a joint cybersecurity information sheet urging public and private sector organizations using artificial intelligence systems and those planning to integrate the technology into their operations to implement best practices and recommendations for AI data security.
In today’s complex and interconnected world, intelligence plays a crucial role in safeguarding the nation, preventing crises and informing policy decisions. The rise of new threats, technological advancements and geopolitical shifts has made intelligence-gathering and analysis more essential than ever. Join the Potomac Officers Club’s 2025 Intel Summit, where the intelligence community’s top leaders will provide insights into the challenges and opportunities facing the IC today and into the future. Register here.
Protecting the data used throughout the development, testing and operation of an AI system will ensure the accuracy and integrity of AI outcomes and help mitigate risks arising from data integrity issues in various stages of AI development and deployment, according to the CSI the National Security Agency published Thursday.
Data Security Risks
The document provides an in-depth examination of three significant areas of data security risks in AI systems, namely data supply chain, maliciously modified data and data drift, and offers general best practices to mitigate such risks and secure sensitive, proprietary or mission-critical data throughout the AI system lifecycle. The best practices include employing digital signatures to authenticate trusted revisions, tracking data provenance and leveraging trusted infrastructure.
Cyber Info Sheet Authors
The document was authored by the NSA’s Artificial Intelligence Security Center, the Cybersecurity and Infrastructure Security Agency, the FBI, the Australian Cyber Security Centre, and the national cybersecurity centers of New Zealand and the United Kingdom.
