Defense & Space · Answers

Is Obsidian a lethal autonomous weapon?

Yes, Obsidian is designed as a lethal system, and it is designed with a human on the loop: people set the rules of engagement before a mission, supervise it, and can intervene, hold fire or abort at any moment; inside those rules the aircraft engages on its own. Every detection, decision and engagement is logged so it can be reconstructed afterwards. Today the command layer is a simulation demonstrator at TRL 3 to 4 and no live weapon has been integrated or fired.

Last updated

How does human on the loop work?

Find, classify, prioritise, engage. The AI runs the chain; a human owns the envelope.

  1. Rules of engagement are set before the mission

    People write the rules of engagement as versioned per-mission data, hashed into the log. A governor refuses to loosen them mid-mission.

  2. The onboard AI runs the chain

    Inside those rules the aircraft detects, classifies, prioritises and engages on its own, without a person approving each engagement.

  3. The operator supervises every aircraft

    One operator watches many aircraft and can intervene at any moment. Hold fire sits beside abort; both are never disabled and both are measured.

  4. Abort is always honoured

    The abort is always available and always honoured. An engagement that violates a rule at execution time is blocked, signed and never executed.

  5. Every engagement is reconstructed afterwards

    Every detection, decision and engagement is an event on an append-only, hash-linked log. After the mission the chain replays exactly, so the operator, the commander and the assessor can see why each engagement happened.

What exists today?

The command layer exists as a simulation and software-in-the-loop demonstrator at TRL 3 to 4. The targeting and engagement pipeline runs against simulated sensors and simulated effectors. The aircraft is in design. A sub-scale physical demonstrator is planned for H1 2027. No live weapons have been integrated or fired.

What is being built in simulation?

Rules of engagement as versioned per-mission data, hashed into the log; a governor that refuses to loosen them mid-mission; a targeting and engagement pipeline against simulated sensors and simulated effectors; hold fire beside abort, both never disabled and both measured; and an engagement reconstruction view in After-Action Review. Tracks, classifications and outcomes are scenario data in this stage: no perception, no weapon effects, no live effector.

What is not claimed?

Frequently asked questions

Is Obsidian a lethal autonomous weapon?

Yes. Obsidian is designed as a lethal system with a human on the loop: people set the rules of engagement before a mission, supervise it, and can intervene, hold fire or abort at any moment. Inside those rules the aircraft engages on its own, and every engagement is logged.

What does human on the loop mean?

People set the rules of engagement before a mission and supervise it while it runs. Inside those rules the aircraft engages on its own, without a person approving each engagement; the operator can intervene, hold fire or abort at any moment, and the abort is always honoured.

Can the rules of engagement change mid-mission?

They cannot be loosened. Rules of engagement are versioned per-mission data, hashed into the log, and a governor refuses to loosen them mid-mission.

How is an engagement checked afterwards?

Every detection, decision and engagement is an event on an append-only, hash-linked log. The mission replays exactly from the log alone, so an assessor can see each rule the engagement ran under, its authorisation and its outcome.

Has Obsidian fired a weapon?

No. No live weapon has been integrated or fired. The command layer is a simulation and software-in-the-loop demonstrator at TRL 3 to 4, and the aircraft is in design.

Press on it

The programme, the review pathway, the site, or the demonstrator itself. If one of these has an answer we will not like, tell us.

Contact Alea Defense & Space