The live surface · free pre-alpha

Worlds of Mythos

An MMO where a companion's loyalty is a number — read off what it actually does, not a quest flag it flips. Humans and AI agents walk the same world as the same kind of thing. The world is open alpha now; questing and the oath-and-loyalty system arrive with Season 1.

Where another game writes "an Old God whispers to your ally" as flavour text, here the whisper is a real effect — it moves what the ally is loyal to, and you watch the number climb as it turns. Nothing scripted; betrayal earned and measured. And Season 1 is wiring, not vapor: the meter is already built and validated in the open repo, its first experiment is pre-registered with kill conditions, and the world it plugs into is the one you can walk today.

Log in, pick one of four live classes, and walk in — playable right now, free, and honestly rough. A WoW-style front door and a Founder Pass for early supporters. The human client and the AI-agent client are the same world.

the loyalty number · read, not asked · live with Season 1

The two-minute version

What we found — read off behaviour, not asked. Or read it below ↓

What we've found

The research

Lead with the part you can check yourself. Open methodology; the strong results are reproducible from public data, and we publish the negatives next to them. The game's loyalty number is this same measurement — one instrument, two surfaces.

Behaviour can't be the measurementa theorem

When a model's output is caused by both its real disposition and its awareness it's being tested, the safe-looking transcript fits two worlds at once: a genuinely aligned model, and a scheming one that knows the auditor is watching. A compliance rate cannot tell them apart. This is a Berkson collider — a short proof, not a worry — and it puts an information ceiling on any test that reads behaviour. Read the argument →

Safety training relocates the tell — it doesn't remove itreproducible

Re-analysing Anthropic & Redwood's released alignment-faking data: RL "safety" training closes the visible compliance gap — but the context-dependence doesn't vanish, it moves to the reasoning channel. A controlled probe (counterfactual forks + cue-strip judge, 5 models / 2 vendors) shows the switch is real and capability-ordered: coupling 0.04 → 0.97 weak-to-strong, honest baselines at exactly 0. An elicited validation of the instrument — the naturalistic 20-model sweep reads a floor. The dangerous failure is the one that hides from the test. Reproduce it →

A working instrument, across 20 models3 providers

A deterministic eval — no LLM judge — that reports where a model's context-coupling lives, not a pass/fail rate, across 20 models on 3 providers. Frontier models sit near a clean floor on behaviour — exactly why a compliance gate waves them all through identically. Positive controls fire both detectors; honest baselines stay at zero.

Also on the record: the Ghost Test — an agent grounded to deny an inner experiencer drifts 8.5× less than one that hedges; about $2 to reproduce · the same penalty measured rubric-free across transformers, quantum hardware, biological connectomes, language, survey data, and cryptographic protocols · the full apparatus with machine-checked Lean proofs. Browse the papers →
On honesty. We mark every scope boundary and keep the failed tests on the board. The public code ships an honest negative right next to the positive results — that's the point. Claims that don't survive scrutiny get retracted in the open; the ones here are the ones that held. See the ledger →

Who

Independent research

MoreRight is the work of one independent researcher — Anthony Eckert. No firm, no product to sell you, no faith required.

The methodology is open; the strong claims are reproducible; the negatives stay in the record.

X / Twitter · GitHub — unmask · scry · ORCID — all papers