Model welfare and moral status

Whether models have interests that matter, and what would follow if they do.

Facts

Lane: Trust and governance
Wave: 10 of 12 (build order, not a date)
Needs: 0 boxes all the way back through hard links
Unlocks: 0 boxes downstream through hard links
Leverage: None
Cluster: Trust plumbing and the agent economy

Evidence as of October 2026

Anthropic says it is "highly uncertain about the potential moral status" of its models, and in August 2025 let Claude end persistently abusive conversations as a low-cost precaution.

01Anthropic: Claude Opus 4 and 4.1 can now end a rare subset of conversations

Around it

Needs · 0

Nothing beyond the detail list. It can start at any time.

Unlocks · 1

Human-agent symbiosisstage · Speculative · helps, not required

Openings researchers' judgement, October 2026 scouting

Startup room 41/100 · rank 57 of 124 Research room 61/100 · rank 29 of 124

Crowding2 Lab overlap3 Small-team fit4 Buyer today1.5 Research openness5

Depth: light
Players: Eleos AI Research; Anthropic welfare programme; Conscium
Wedge: Valid welfare indicators, including reliability of model self-reports and introspection. research

01Eleos: AI welfare research priorities