Resolves the same as the Metaculus question:

https://www.metaculus.com/c/ai-warning-signs/31331/intl-aggression-over-ai-before-2030/

Resolution Criteria

This question resolves as YES if the following occurs before January 1, 2030, and resolves as NO otherwise:

A nation state commits an act of aggression against another nation state OR threatens the same, explicitly due to the target nation state's real or supposed AI capability advances.

Fine Print

1. An "act of aggression" is here defined by United Nations General Assembly Resolution 3314.

2. A "nation state's real or supposed AI capability advances" are here defined as improvements in artificial intelligence capabilities, including but not limited to: development or possession of powerful AI systems, major technical breakthroughs in the field of AI, acquisition of significant compute for AI, or mere intent to enhance AI development capacity. The advances may be real or supposed, and don't have to be directly attributable to or controlled by the nation state.

3. For an act or threat to count, the reasoning linking the target nation state's real or supposed AI capability advances to the threat or act must be made explicit.

4. Implicit threats are not on their own sufficient for a YES resolution. An explicit verbal or written threat may be sufficient for a YES resolution if the other criteria are met.

5. For a threat to count, it can be direct ("we will attack") or conditional ("do X or we will attack"), but not purely hypothetical ("if we thought they could do X, we would attack") or purely defensive against an act of aggression ("if they attack us, we will be forced to respond").

6. For a threat to count, it must come from someone who can credibly make such a threat. (e.g. heads of state, senior cabinet members, or military leadership. Not opposition politicians, lower-ranking diplomats, media, etc.)