‘We really do earnestly believe AI could kill all humans’: if AI labs are so worried about AI doom, why don’t they stop?

Original reporting

Read the source story

The Conversation‘We really do earnestly believe AI could kill all humans’: if AI labs are so worried about AI doom, why don’t they stop?11 Sept 2026Open article ↗

Current understanding

One reported view is that some AI labs earnestly believe advanced AI could kill all humans, according to a publisher. [1]

The headline frames a question about why, if those labs are so worried about AI doom, they do not stop their work. [1]

AI-generated from linked publisher headlines. GDELT-discovered reporting is credited to the original publisher; AtlasWire does not retrieve article bodies. Generated 11 Sept 2026, 05:23 UTC.

Impact assessment

10/100

Impact assessment

routine human impact

The score is supported by one ownership group and is not independently corroborated.

Score breakdown

What shaped this score

Each dimension is scored separately out of 100.

Loss of life and injury10/100

People affectedNot yet evidenced

Living conditionsNot yet evidenced

Ongoing riskNot yet evidenced

Duration and relief accessNot yet evidenced

Wider societal impact1/100

The strongest evidenced dimension establishes the base score. Separate human-impact dimensions can add a bounded share of the remaining headroom without double-counting the same harm.

What the reporting supports

Reported deathsDeaths reported without a reliable countkill
Scientific impact1 scientific impactAI could kill all humans