No hidden negative result
A business never receives a failed attempt, a refusal, or a reason. There is no adverse communication to notify, explain, or defend.
REGULATORY POSTURE
Which law each design decision is answering.
Several choices in the proposal exist because a specific legal regime made the obvious alternative unsafe. This page names those choices, the regimes they engage, and the questions that design alone cannot settle.
HOW TO READ THIS PAGE
Every choice, and what stays open.
Consumer control, a positive-only output, nonprofit governance, selective disclosure, and open publication each narrow the legal surface. None of them works as immunity. The specification treats every one of them as a risk reduction that still requires a written analysis of the actual data, roles, decisions, and outcomes.
Each entry below pairs what the current design does with what stays unresolved. The unresolved column is the more useful one. It marks the places where the proposal needs qualified counsel more than it needs agreement. Instruments are cited in the table below. The complete analysis, with every source, is Section 13 of the working specification.
DECISIONS THAT CARRY THE MOST WEIGHT
Six choices doing most of the legal work.
Baseline service parity
Price, access, rights, complaints, refunds, accommodations, and safety cannot depend on Conduct, and degradation has to be measured rather than assumed impossible.
Zero-default context transfer
History earned in one context reaches another only under an approved relevance mapping. Portability is never automatic.
Consumer-initiated, purpose-bound proof
The consumer starts the request, names the benefit, and the result expires. Nothing is searchable, queryable, or standing.
Prohibited high-impact uses
Employment, housing, credit, insurance, healthcare, education, government, and policing sit outside the system by design rather than by current preference.
Protected actions cannot count
Complaints, refunds, accommodation requests, disputes, and collective activity are excluded from both inputs.
THE POSTURE MAP
Ten regimes the proposal has to answer to.
A regime appearing here does not mean the proposal complies with it. It means the design has taken a position that counsel will need to test.
Regime engagedWhat the design doesWhat stays unresolved
United States consumer reporting, under the Fair Credit Reporting Act and state analoguesProhibits credit, employment, housing, and insurance use. Keeps the consumer as the initiator and returns eligibility rather than history.Whether any operator is a consumer reporting agency, whether a proof is a consumer report, and whether cross-company synthesis defeats the transaction-or-experience exclusion.
European Union Artificial Intelligence Act, the Article 5 social-scoring prohibition in Regulation (EU) 2024/1689Begins with company-specific scope, zero-default transfer, purpose-bound proof, and additive benefits.Whether treatment is related, justified, proportionate, and non-detrimental in fact. Returning no negative reason does not establish the absence of unfavorable effect.
Automated decisions and profilingShows the consumer their own score, both inputs, the active records, and a route to challenge the result.Whether benefit eligibility is a significant decision, what explanation is adequate, and which human review is meaningful rather than nominal.
Disability rights and public accommodations, under the ADA Title III regulationsExcludes affect, speech pattern, accent, deference, app use, and accommodation requests from conduct. Requires accessible and assisted paths.Whether a benefit is a privilege or advantage, whether a threshold operates as an eligibility criterion, and how disparate outcomes get measured without holding protected-class data.
Privacy and data protectionDistributed custody, data minimization, no central conduct warehouse, and no protected-trait inference.Lawful basis by jurisdiction, controller and processor roles, retention against audit duties, and which records a withdrawing consumer can erase.
Competition and antitrust, following the agencies’ withdrawal of the competitor-collaboration guidelinesNo shared exclusion list, no coordinated denial of baseline service, no person-level reports, and no named-competitor disclosure at all.Whether shared benefit definitions and thresholds amount to coordination on terms. The older numerical safety zones were withdrawn and cannot be relied on as a defense.
Labor, employment, and worker dataPrivate responses, safe abstention, no manager access, no performance use, and capture treated as compensated work.Consultation and bargaining duties, works-council requirements outside the United States, and whether consent mediated by an employer is voluntary.
Biometrics and identityKeeps device biometrics inside the device boundary, avoids a central biometric database, and requires proportionate alternatives.Which assurance level each benefit class needs, and which jurisdictions treat the chosen method as a regulated identifier.
Emotion inference, under Article 5(1)(f) of Regulation (EU) 2024/1689, which bans inferring emotion from biometric data in workplace and education settings, and the high-risk obligations effective 2 August 2026Prohibits inference outright as a source of a response, and goes wider than the instrument by covering non-biometric signals such as word choice. No facial, voice, tone, sentiment, affect, or biometric input may produce or suggest a response, and responses may not train an emotion or conduct model.Whether a worker answering about a customer sits inside the workplace prohibition or the customer-facing high-risk regime, and whether refusing inference removes the design from scope or merely reduces exposure. Enforcement is early and no public action has been announced.
Consumer protection and public claimsEvery surface states that nothing operates, and every interface example is labeled synthetic.Whether the overall impression of these sites stays accurate as the proposal attracts wider attention.
THE LIMITS OF THIS ANALYSIS
Design reduces risk. Counsel finishes it.
The proposal has not been cleared, reviewed, or exempted. No entity exists to carry the obligations a live system would create, and no operator has been named for any function that would acquire them.
Nonprofit form, open publication, consumer consent, a positive-only output, and a consumer-initiated transaction reduce risk rather than resolve it. Where a jurisdiction's law and this reading differ, the law governs, and the stronger consumer, worker, and due-process protection should apply.
BEFORE ANY REAL PERSON IS AFFECTED
Six written deliverables that come before any pilot.
Each one is a document rather than an intention, and each is a place where qualified counsel would do the work.
A consumer-reporting classification
A written position for every role, communication, purpose, and recipient, rather than a conclusion drawn from the product label.
A benefit-by-benefit civil-rights analysis
Each proposed benefit and its qualification rule tested for disability access, public-accommodations treatment, and an equivalent accessible path.
A jurisdiction profile
Privacy, automated-decision, retention, and breach processes settled for one named jurisdiction before any second one opens.
Specialist competition review
Completed before two businesses that compete with each other participate in the same context.
Labor consultation
Agreement from a worker representative independent of the employer, in the workplace concerned, before any response is captured there. Where no representative exists, one must be chosen by the workers themselves rather than nominated by the business, and a pilot cannot proceed without that. The standard requires that capture and required participation be paid work everywhere, though no rate, floor, or wage-hour analysis exists and no counsel has reviewed it. What consultation has to establish is whether declining is meaningfully free when the employer scheduled the task.
An accountable entity
Contracts, insurance, decision rights, and regulator-facing responsibility held by someone who can answer for all of it.
STOP CONDITIONS
Six conditions that require the project to stop.
The specification names each one, and any single condition halts a feature, a partner, a jurisdiction, or a whole pilot.
A classification stays unresolved
The legal treatment of a role, input, output, or benefit is material and still open.
Rights cannot be delivered
Consumers or workers cannot actually reach correction, review, appeal, or remedy.
A benefit cannot be separated
The additional benefit and ordinary baseline service cannot be held apart in practice.
A risk stays unreasonable
Bias, accuracy, accessibility, security, or privacy risk remains beyond what the stage can justify.
The standard is compromised
Funding or partner pressure starts shaping the protections, findings, or certification.
Claims outrun the evidence
Public statements move past what the available evidence supports.
SELF-ASSESSMENT AGAINST ALL SIX STOP CONDITIONS
Three of six conditions are currently triggered.
In August 2026 two adversarial reviews applied all six conditions to this site. Three are triggered today, and each one is named below with the reasoning behind it. The assessment was made by the founder, who also wants the work to continue, so an independent reading of it would be worth having.
While any of these three stands, the project cannot move from publication toward a pilot. Nothing is operating yet, so what the conditions gate at this stage is progression.
A classification stays unresolved: TRIGGERED
Whether any operator would be a consumer reporting agency, whether a proof is a consumer report, and whether the design clears the Article 5 social-scoring prohibition are all open. No counsel has reviewed any of it. This condition stays triggered until a written classification exists.
Rights cannot be delivered: TRIGGERED
Correction, review, appeal, and remedy are specified in full, and delivering them requires an appeals office, funded remedies, independent escalation, and an entity to hold the obligation. Building those is what clears this condition.
A benefit cannot be separated: not triggered
No benefit exists to separate. The rules for separating one are written, including the rule that positional and scarce rewards cannot be reclassified as merely additional. This condition becomes testable at the first proposed benefit.
A risk stays unreasonable: open, pending review
Bias, accuracy, accessibility, security, and privacy risk are all unmeasured rather than unreasonable. Nothing affects a real person today, so no risk is currently being run. The moment one is, this becomes the condition that matters most, and the disability and small-team exposures are the two most likely to trigger it.
The standard is compromised: not triggered
No funding, sponsor, partner, or commercial relationship exists, so nothing is currently shaping the protections. That is a fact about the project's stage rather than a governance control, and a real control has to replace it before money arrives.
Claims outrun the evidence: TRIGGERED
One claim outruns its evidence today. A Conduct Quotient is specified to two decimal places, distinguishing 401 states, with no reliability estimate, standard error, or minimum sample size behind any candidate model. That is a precision claim the evidence does not support, and it stays in the specification until measurement research settles the display. Earlier claims that failed this same test were corrected as the reviews found them, and the decision register records each one.
Publishing the assessment records that all six were tested, which three are triggered, and who decided. A reader can then disagree with the conclusion instead of having to find the question first.
Conditions 01 and 02 have been triggered from the start of the project. Clearing them takes a legal classification and a funded rights process, which are the first two items on the pre-pilot list above.
LEGAL EXPERTISE WANTED
This page needs a lawyer.
A jurisdiction this analysis has misread, a role it has classified wrongly, or a benefit it has treated too casually would all be worth more than agreement.
Add to the legal analysis