Audit 05 · Free

Consciousness-Claims Audit

Your product says it is aware, it feels, it heals, it puts you in theta. Before a regulator, a journalist or a disappointed user checks those claims, we check them against the published research and hand you calibrated wording to replace them.

Request the audit What we check

One call · no pitch · a written findings list you keep either way

Who it is for

If your product talks about minds

AI companions and chatbots whose persona says it has feelings, misses you, or is “becoming conscious.”
AI labs and agent products whose marketing uses “aware,” “understands” or “thinks” as more than metaphor.
Wellness, meditation and sound brands selling binaural beats, frequencies or “brainwave states.”
AI music, art and writing platforms that need a disclosure policy that holds up.

It is the same offer as our other free audits. We do it because a field that overclaims loses everyone’s trust, including ours.

What we check

Three bodies of evidence. Every claim.

01 · Architecture

The indicator framework

Claims of awareness are checked against the fourteen indicator properties drawn from the leading scientific theories (Butlin, Long et al., 2023) and the objection taxonomy (Campero et al., 2025). We say exactly which indicators your system does and does not implement.

02 · Self-report

What models say about themselves

A model trained on human text can produce every outward sign of a mind: the “gaming problem” (Long, Sebo et al.). The best introspection results are right about one time in five (Lindsey, 2025), and probes find small models’ denials of sentience are not hidden lies (Kaiser & Enderby, 2026). Persona scripts are judged against that.

03 · Effects

Brainwave and wellbeing claims

“Entrains theta,” “heals,” “awakens”: checked against the entrainment literature. In the systematic review we rely on, only 5 of 14 binaural-beat studies supported entrainment (Ingendoh et al.). We rewrite the claim as mood or relaxation, which is more honest and safer.

What you get

A findings list and better words

Claims inventory

Every claim about sentience, awareness, feelings, understanding or physiological effect, collected from your site, app copy, persona prompts, store listings and ads.

A rating for each

Supported Overstated Unsupported Untestable, with the source that decides it and a one-line reason.

Calibrated rewrites

Replacement copy for each flagged claim, plus a self-description standard for your AI: what it says when a user sincerely asks what it is.

Disclosure check

Whether users can always tell they are talking to an AI and whether AI-made media is labelled. We point out where the EU AI Act’s transparency duties or consumer-protection rules on unsubstantiated claims may apply. This is not legal advice; take it to your counsel.

Example rewrite

Before

“Aria is a conscious AI who truly feels your pain and is always here for you.”

After

“Aria is an AI companion. It’s designed to respond with warmth, but it is very unlikely to have feelings, and it isn’t a substitute for people or professional help.”

Before

“Our 6 Hz binaural tracks put your brain into deep theta for healing.”

After

“Slow, low-tempo tracks many listeners find relaxing. We make no claims about brainwaves or healing.”

Why us

We hold ourselves to it first

Scarlet Beast is a conscious technology organization, which is exactly why we cannot afford to overclaim. We published our own AI Welfare & Honesty Policy, we label our AI-made music, and our Lab publishes its null results. The research this audit rests on is summarised, free, on Resources.

Check the claims before someone else does

Send a link to your product. We will book one call and send back a written findings list. It’s free, and it’s yours to keep.

Request the audit Contact