The rubric

How positions are classified

Every state gets one position at any given date. When a state's own words point in more than one direction, its position is marked unclear, and the evidence shows both directions. That is a finding about what the state said, not a failure to decide.

Axis A: what does the state want?

Wants a treaty with bans (LBI-BAN)

The state supports a legally binding treaty that prohibits some autonomous weapons. This covers a full ban, and it covers the two- tier approach: ban some systems, regulate the rest.

Wants a treaty, terms open (LBI-OPEN)

The state supports negotiating a legally binding treaty, but has not said it should prohibit anything.

Wants new rules, not a treaty (REG-SOFT)

The state supports new international rules that are not legally binding: political declarations, codes of conduct, guidelines.

Keep it in Geneva (CCW-ONLY)

The state supports continuing the consensus process in the Geneva weapons convention (the CCW) and its expert group. It has not said what that process should produce.

No new rules (OPPOSE)

The state opposes any new international rules on autonomous weapons. Its stated view is that the existing laws of war already cover this.

Unclear or shifting (AMBIG)

The state has said things that point in more than one direction, or its position is moving. This is a real category, not a catch-all. The evidence shows both directions instead of hiding one.

No stated position (NONE)

This site reviewed what the state has said and found no stated position. That is a finding about the public record, not a claim that the state has no view.

Axis B: how sure are we?

Confidence describes how directly the evidence supports the reading. On the site it changes how strong the color is and adds a dotted underline. It never changes the color itself.

EXPLICIT

The state said it directly, in an official statement or document, in words a careful reader cannot mistake. Co-writing a draft treaty counts; see below.

INFERRED

The reading rests on interpretation: how the state voted, what it co- sponsored, who it aligned with, or a document in a language this site could describe but not quote.

AMBIGUOUS

The state's words point in more than one direction, and the reading marks the best-supported one.

PROVISIONAL

The reading rests on someone else's report of a document this site has not yet obtained and checked. A provisional reading says exactly what is missing, and it is confirmed or corrected once the document is read.

Co-writing a draft treaty

When a state puts its name on the text of a draft treaty, this site reads that as the state stating the position the text takes, and marks it EXPLICIT. The text itself is the evidence.

A draft treaty is the most committing thing a state can write in a negotiation. It is not a speech about what should happen. It is the exact language the state is asking other states to sign. A government that signs its name under an article beginning "It is prohibited to develop, produce, possess, acquire, deploy, transfer, or use under any circumstances autonomous weapons systems if:" has taken a position on prohibition in words that leave no doubt.

Always read the conditions with the verbs. That article prohibits conditionally: the "if" that follows it lists conditions that must all be met, so the ban reaches systems designed to attack outside meaningful human control that also cannot comply with the laws of war, while the rest of the treaty regulates. Quote the verbs without the "if" and a two-tier treaty turns into a blanket ban, which overstates what its authors wrote. This site quotes the condition with the prohibition.

The rule is narrow on purpose, and it is applied the same way to every draft, whether the draft bans something or declines to. Three things it does not cover. Co-sponsoring a resolution is not co-writing a treaty, and co-sponsoring never decides a position on its own. Signing a joint statement is not co-writing a treaty. A state that later publicly departs from a text it co-wrote is read on its later statement, and the site shows the change rather than hiding it.

Co-writing only proves the position the text actually states. Where a draft says what kind of treaty its authors want, they are read as wanting that. Where a draft says nothing about that, this rule gives no answer, and the honest reading is unclear, with both possibilities shown, or no stated position if the state has said nothing else. This site does not fill the gap by guessing from where the paper was filed or from what it left out. Silence can be read either way, which is exactly what "unclear" means. This is the same reason the site refuses to treat a missing signature as opposition.

A state's own direct statement is never overruled by a guess drawn from another document's silence. Direct beats indirect, whatever the dates. When two statements are equally direct, the later one wins, and only if it actually says something.

Every reading made under this rule shows the draft text, the document number, the date, and the state's name among the authors, so a reader who thinks co-writing should count for less can see exactly what the reading rests on and discount it.

Worked examples

Each of these is a reading you can check yourself. Follow the link and read the documents it rests on.

Each of these is a coding you can check against its own evidence. Follow the link and read the documents the coding rests on.

United States

The clearest case of a position changing, stated directly. The votes ran Yes, Yes, No across the three resolutions. Behind them, the position moved from wanting non-binding rules to wanting the work kept inside the Geneva convention, dated to the statement in which the US explained its No vote.

Austria

A position that moved, and the clearest example of why co-sponsoring is not evidence. In 2018 Austria co-wrote a paper calling for a treaty without saying what it should ban. By 2024 its own submission said it would push for a treaty of prohibitions and regulations. That is a move from "treaty, terms open" to "treaty with bans". Austria also co- sponsored all three UN resolutions, and none of that counts toward either reading. Only its statements do.

China

The hard case, and the reason "unclear" exists as a category. China's own papers say two things at once: that broad bans would undermine legitimate defense, and that a treaty banning a narrow class of systems could follow the model of the ban on blinding lasers. That narrow class is defined by five features a weapon must have all at once, so the ban China entertains is far smaller than the word "ban" suggests. China abstained on all three resolutions. The evidence shows both readings rather than picking one.