Sam Altman: Amen.dment to Work with Department of Defense to clarify that AI will not be used to engage in surveillance that violates the privacy rights of American citizens

March 3: OpenAI CEO Sam Altman retweeted an internal memo stating, “We’ve collaborated with the U.S. Department of Defense (DoD) to clarify our principles via updates to our agreement. We’ll add the following clause as a supplement to existing terms: As required by applicable U.S. law—including the Fourth Amendment to the U.S. Constitution, the 1947 National Security Act, and the 1978 Foreign Intelligence Surveillance Act (FISA)—AI systems shall not be intentionally used for domestic surveillance of U.S. citizens or nationals. For clarity, the DoD understands this restriction prohibits intentional tracking, monitoring, or surveillance of U.S. citizens/nationals, including through the acquisition or use of commercially available personal or identifiable information.” Protecting U.S. citizens’ civil liberties is a top priority. Given widespread external interest in this issue, we’re clarifying this point—including regarding commercially obtained information—consistent with our iterative deployment approach, as we’ll continue learning and refining. I view this as a critical amendment; the DoD has also confirmed our services won’t be utilized by DoD intelligence agencies (e.g., the NSA). To be clear, we aim to operate via democratic processes. Technology remains immature in many areas, and the required security trade-offs are still unclear in many cases. During weekend discussions, I reiterated that Anthropic should not be designated as SCR and expressed hope the DoD will extend to them terms similar to those in our agreement.