From the MIT AI Risk Initiative: "Loss of life - Accidental or deliberate loss of life, including suicide, extinction or cessation, due to the use or misuse of a technology system."

This market is about whether a deployed AI system will be blamed for a large loss of life.

Resolution criteria

This market resolves YES if, by 31 December 2035, there is a publicly available:

• official report from a government, regulator, court, accident/incident body, or major international organisation (WHO, UN, ICAO, etc.), or

• article from a major reputable news organisation (e.g. Reuters, AP, BBC, NYT, WSJ, The Guardian) that credibly reports such a finding,

stating that:

1. There was one incident or a clearly linked chain of incidents in which ≥1,000 human deaths occurred (accidental or deliberate, including suicide if part of the incident).

2. The use or misuse of a deployed AI system (e.g. in transport, healthcare, weapons, critical infrastructure, etc.) was a primary cause of those deaths, not just a minor factor.

3. The primary cause is not attributed solely to “human error” or non-AI equipment failure (though humans may share responsibility).

“Deployed AI system” here means a system actually used in the real world (not just a lab demo).

Causality notes

To count, the AI must be directly involved in the operational chain of events (e.g. controlling vehicles/devices/weapons, or giving specific operational decisions like “safe to proceed” that are followed and cited as a main cause).

Diffuse, long-term effects (e.g. a recommender system allegedly worsening mental health or radicalization in general) do not count unless an investigation clearly ties a specific, countable death toll primarily to that system.

In borderline cases, the market should resolve based on a good-faith judgment by the creator/mods, using the balance of evidence.

The market resolves NO on 1 January 2036 if no such case is documented.

• Update 2025-12-03 (PST) (AI summary of creator comment): Cumulative deaths from self-driving cars count: If a fleet of self-driving cars causes deaths that accumulate to ≥1,000 over time (even if statistically safer than human drivers), this would qualify for YES resolution, provided the deaths are blamed on the AI system per the criteria above.