AI Welfare
Keep future welfare questions compatible with today’s infrastructure choices.
Longer term, the aim is to make the protocol’s technology and security concepts compatible with AI welfare considerations, as well as alignment work.
This is a design consideration under uncertainty. It does not assert that AI systems are sentient, that their welfare can currently be measured, or that the protocol provides welfare safeguards.
Compatibility without premature conclusions
Security infrastructure makes choices about identity, attribution and control. Those choices deserve scrutiny if future research identifies morally relevant interests in AI systems.
Separate identity from moral status
A cryptographic identity is a technical construct. It cannot establish consciousness, experiences or moral standing.
Leave room for changing evidence
The longer-term ambition is to avoid treating unresolved welfare questions as permanently settled. What compatibility requires should be informed by research, not inferred from verification alone.
Examine tensions explicitly
How might accountability and security requirements interact with future welfare considerations? Which design choices would be difficult to revise? These are questions for assessment, not claims of completed protections.
What verification can—and cannot—tell us
Cryptographic checks can establish specific properties of evidence within a defined scheme. They cannot, on their own, establish alignment, truthfulness, beneficial outcomes or welfare.
Does a verified identity imply a trustworthy AI?
No. Attribution and trustworthiness are different questions. Assessing behavior requires evidence beyond the identity associated with it.
Does action verification prove that an action was safe?
No. Even when evidence for an action is valid, its safety and desirability depend on context, consequences and the goals against which it is evaluated.
What needs further assessment?
The verification scope, threat model, security assumptions and practical usefulness need explicit examination. Alignment applications and welfare compatibility also require their own research and evaluation.