OpenAI's Approach to AI and National Security

The US White House has announced a National Security Memorandum (NSM) to strengthen national security using AI and to establish guardrails. OpenAI agrees that securing semiconductors, power, and data center capacity is essential for maintaining US AI leadership, and views this NSM as an important step forward.

OpenAI believes AI is at an inflection point where it can either strengthen or weaken democratic values. Therefore, it emphasizes that democratic AI leadership must be secured based on values such as freedom, fairness, and respect for human rights.

OpenAI is currently collaborating in the following areas:

• DARPA: Cyber defense and protection of critical networks
• USAID: Reducing administrative workload
• Los Alamos National Laboratory: Bioscience research partnership

To evaluate national security partnerships, OpenAI applies the following four core values:

• Democratic Values: Expanding technology access and maximizing economic and social benefits
• Safety: Rigorous evaluation to mitigate risks and protect human rights
• Responsibility: Use for the public good and prevention of violations of fundamental rights
• Accountability: Oversight of government and security-related applications and adherence to ethical standards

Going forward, OpenAI is keeping open the possibility of applying its technology across various fields such as scientific research, logistics improvement, translation and summarization, and mitigation of civilian harm, with all work set to undergo rigorous internal review.