I use the word monopoly to draw your attention to other matters.
Open AI has a contract with the United States Government. What are we doing here, commingling data?
Protections for Americans’ Personal Information
Americans’ personal information is protected under a combination of U.S. constitutional law, federal statutes, and contractual agreements when AI systems like those from OpenAI are involved in government activities:
- The Fourth Amendment protects against unreasonable searches and seizures.
- The Foreign Intelligence Surveillance Act (FISA) and Executive Order 12333 regulate how intelligence agencies can collect and use data, especially involving U.S. persons.
- OpenAI’s contract with the U.S. Department of Defense explicitly states that its AI systems shall not be used for unconstrained monitoring of U.S. persons’ private information.
- Recent amendments to the OpenAI–Pentagon agreement prohibit “deliberate tracking, surveillance, or monitoring of U.S. persons or nationals,” including through commercially acquired data.
However, critics argue that terms like “deliberate” and “intentional” leave room for incidental data collection, which has historically been exploited under existing legal loopholes such as the data broker loophole and backdoor search loophole.
Is This Partnership Based on a Treaty?
No, the OpenAI – South Korea cyber defense partnership is not based on an international treaty. It is a bilateral cooperation agreement between a private company and a foreign government.
The only existing binding international AI treaty is the Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, signed by the U.S. in September 2024. While this treaty promotes responsible AI use and includes cybersecurity provisions, it exempts national defense-related AI systems from many of its obligations, provided they respect human rights and democratic processes.
Key Safeguards and Oversight Mechanisms
OpenAI emphasizes a multi-layered protection strategy in its government partnerships:
- Cloud-only deployment: Ensures no local model installations that could bypass safety controls.
- Human-in-the-loop: Cleared OpenAI personnel monitor usage to enforce compliance.
- Safety stack control: OpenAI retains full authority over model guardrails and can terminate access if terms are violated.
- Data residency options: Particularly relevant in partnerships like South Korea’s, where data can remain within national borders or not be stored at all.
Despite these measures, legal experts caution that contractual language alone may not prevent misuse, especially if government agencies interpret existing laws broadly to justify surveillance activities.
I am not asking AI this question – I am asking you. Is there any government oversight to ensure the safety of American data?
We need a firewall; tested by global white-hats. As you have made this a global matter.
The big difference between now and heretofore is the nature of Artificial Intelligence; totally new.