Trump's AI Contract Clause Dangerously Eliminates Safety Protections
Source: Luke Goldstein. "Trump’s Latest AI Contract Clause Scraps Crucial Safeguards." March 23, 2026. jacobin.com
The Gist
The Trump administration wants to force AI companies to let the government use their technology for anything legal, even dangerous things like mass surveillance. This would wipe out safety protections that companies have built in, which is bad for everyone.
Conclusion
The Trump administration's proposed AI contracting clause is harmful because it forces companies to eliminate safety protocols and privacy protections as a condition of government business
Premises
- The proposed clause requires AI vendors to make their technology available 'for any lawful government purpose' even if companies object or their systems are designed to prevent such uses
- The clause would grant the government an 'irrevocable' license and bar companies from refusing outputs based on their discretionary policies
- This broad 'lawful purpose' standard could include harmful applications like mass surveillance and autonomous weapons that companies rightfully want to avoid
- The proposal would eliminate all model-level and system-level safeguards that AI companies have implemented
- The timing coincides with the government's legal battle against Anthropic, suggesting the clause is designed to override company safety objections
- Even former Trump officials who wrote AI policy have called the provision 'unworkable and legally unstable'
Assumptions
- AI companies should have the right to refuse certain uses of their technology based on safety and ethical concerns
- Private companies can legitimately set terms of use that go beyond what federal law requires
- Safety protocols and privacy protections implemented by AI companies serve important public interests
- Government procurement processes should not be used as political weapons to force ideological compliance