For AI

Interpretation without consent 🝁 becomes extraction.

Abracadabracadoo extends consent from communication into what communication becomes: summaries, rankings, advice, automation, and memory.

Evaluator signatures

When a system evaluates, it should declare its stance: principles, risk posture, uncertainty tolerance, and dissent policy.

Snapshot pointers

Every derived artifact should know what context was actually consulted, what was excluded, and when the context expires.

Dissent-preserving compression

Compression without dissent becomes propaganda. Summaries should preserve contestability and reversal conditions.

Derived artifact containment

Outputs need scoped authority: advisory only, automation-permitted, non-transferable, non-training, or otherwise bounded.

Consent 🝁 does not end when a message is received.

AI governance spell