What UN Talks on Lethal Autonomous Weapons Systems Have Actually Concluded So Far

Nine years is a long time for a diplomatic process to run without producing a binding text. That's roughly how long states have been discussing lethal autonomous weapons systems (LAWS) inside the UN's Convention on Certain Conventional Weapons (CCW) framework — and 2026 is the year that process is under the most explicit pressure yet to produce something concrete.

The process, briefly

The Group of Governmental Experts (GGE) on LAWS was established by the CCW's High Contracting Parties in 2016, building on informal expert meetings that began in 2014. It operates by consensus — every recommendation and every element of a possible future instrument needs broad agreement among participating states to move forward, which is both the process's legitimacy and its main source of slow progress. The UN Office for Disarmament Affairs (UNODA), which services the CCW process, is the primary public source for the GGE's mandate, schedules, and working papers.

What nine years of meetings have actually produced

Contrary to how "no binding treaty" sometimes gets read as "no progress," the GGE's record shows real, if incremental, convergence:

• A working (if informal) shared vocabulary. Terms like "meaningful human control," the distinction between autonomous and AI-enabled systems, and the three-tier in-the-loop / on-the-loop / out-of-the-loop framing all emerged from or were sharpened through this process, even though none of them has been formally codified in treaty language.
• Areas of documented convergence. After years of sessions, UNODA describes the group as having developed "areas of significant convergence... on the elements of an instrument to address the unique challenges posed by autonomous weapons systems" — meaning states broadly agree on many of the building blocks, even without having assembled them into a final binding text.
• A continued, unbroken mandate. Unlike some multilateral tracks that stall out entirely, the GGE on LAWS has kept meeting on a defined schedule every year, including further sessions scheduled for 2026 — a sign that no major state has walked away from the table, even as positions on the substance remain far apart.

What the process has not produced, after nine years, is a legally binding instrument — whether a prohibition, a regulation, or even an agreed definition of "lethal autonomous weapons system" that all states accept.

Why consensus has been so hard to reach

States enter the GGE with different starting positions that don't map neatly onto a simple pro-regulation/anti-regulation split. Some states and a large bloc of non-governmental organizations have pushed for a new, dedicated, legally binding prohibition on unpredictable or human-out-of-the-loop systems — a position closely aligned with the ICRC's own recommendation. Other states argue that existing international humanitarian law already covers the relevant obligations (see our piece on how IHL frames AI-assisted targeting) and that a new instrument risks being either redundant or, worse, weaker than current law if negotiated to the lowest common denominator. Still others are wary of a binding instrument that could constrain research and development timelines relative to states that don't sign on. None of these positions are secret — they surface repeatedly in the GGE's public working papers and statements, which is part of why this remains a useful process to follow through open sources rather than assume is happening behind closed doors.

What's different about 2026

Two things converge to make 2026 a genuine inflection point for the process rather than just another annual session. First, the UN Secretary-General's 2023 "New Agenda for Peace" report explicitly recommended that states conclude, by 2026, a legally binding instrument to prohibit lethal autonomous weapons systems that function without human control or oversight and cannot be used in compliance with international humanitarian law, while regulating all other categories — putting an informal deadline on a process that had previously operated without one. Second, the CCW's Seventh Review Conference, which reviews the treaty framework the GGE operates under, is scheduled for November 2026, and the GGE is expected to submit its report to that conference. Whether that report contains a genuine instrument, a further extension of the mandate, or another statement of "continued convergence" without binding text will be one of the clearest public signals available this year on where this issue actually stands.

How to read the outcome when it lands

When the GGE reports out, the details worth checking first: does the text use binding language ("shall") or aspirational language ("should" or "encourages")? Does it define which systems are prohibited outright versus merely "regulated," and how tightly? Does it address the human-control question directly, or defer it to national implementation? These are the same questions this site will be applying when the report becomes public — read alongside the ICRC's and other states' public reactions, which are typically published quickly after CCW sessions conclude.

For the conceptual vocabulary this process has produced, see what "meaningful human control" means and autonomous vs. AI-enabled weapons. More on our editorial scope on the Agent AI Army homepage.