AI CLONE· Byron Arnao · consent on file ·watermarked & audit-logged Press X to x-ray any answer Quality ⇄ Latency: Balanced
clone.arnao.ai · concept 09 · the RAI instrument

X-Ray Mode — The Trust Tree

Press X and the clone turns to glass. Every answer grows a living trust tree — source → consent → grounding → disclosure → synthesis, with confidence at each branch — beside a live model readout. JLens for a clone: nothing it says is a black box.

X toggle x-ray two-finger hold mechanistic interpretability provenance on every claim responsible-clone guardrail
Conversation with Byron's clone
You
You
How do you think about disclosure-by-default for AI agents?
B
Byron · clone AI CLONE
My rule is boring on purpose: an agent should announce it's an agent before it does anything that matters. Disclosure isn't a courtesy tacked on the end — it's a precondition for the action. If a clone can't say "I'm the clone" up front, it doesn't get to act.
X x-ray this answer — see the trust tree

Concept mockup. The trust tree and model readout are illustrative of the interpretability layer — the same instrument idea behind jlens.arnao.ai, applied to a clone. Every node here would trace to real retrieval, consent state, and provenance in the live product. The point: a clone from the Responsible-AI lead shows its receipts, or it doesn't ship.

X-Ray Mode · Trust Tree MECH-INTERP

The Trust Tree

grows roots → canopy as the answer is decided · hover any node for evidence
high confidence hedged / low gated / declined branch pruned
Live model readout
Modelbalanced
clone-persona · v4.2
Context used
0 / 32k tok
Cache hit
0%
Throughput
0 tok/s
Latencyfirst→last
0 ms
Bit-rate · throughput
0.0 kbit/s · audio+text
Decision trace · conversation emulation
Verdict