OpenAI drops the ban on "military and warfare"
On 10 January 2024 OpenAI rewrote its usage policies. Until then the page barred "activity that has high risk of physical harm", naming "weapons development" and "military and warfare". The blanket prohibition on military use disappeared.
Why it matters
A line the laboratories had treated as self-evident turned out to be an editorial sentence that could be removed without announcement. From then on the boundary between a model and an army was set by contract rather than by company policy, which is why the later records in this line are about contracts and courts.
The change was not announced. Sam Biddle found it and described it on 12 January 2024. The changelog on the new page said OpenAI had updated the policies "to be clearer and provide more service-specific guidance", and a note at the top called them "more readable". The new text kept an injunction not to use the service to "harm yourself or others", with "develop or use weapons" as an example. The prohibition on weapons survived; the general prohibition on military use did not, and neither "military" nor "warfare" appears on the new page. Spokesperson Niko Felix put it this way: "We aimed to create a set of universal principles that are both easy to remember and apply, especially as our tools are now globally used", adding that there are national security use cases that align with the company's mission. The wording before and after is quoted from the page itself, in archived copies of 10 January 2024; the spokesperson's words are as The Intercept gives them.