Trust and safety

How we govern our systems, who oversees them, and how to raise a concern.

Approved by the Board for implementation · 20 November 2026

How we govern our systems, who oversees them, and how to raise a concern. Much of what follows is proposed, not yet adopted; each item says which.

Status

In place today: the Board and Delaware law. Approved by the Board for implementation (20 November 2026): Founding Charter, Bylaws, AI Constitution, Contributor Rights Charter. Still to be formed or filled: 4 oversight bodies and roles. CES pilot: Wave 1 planned for 27 November 2026. CES policies published: 0 of 5.

Raise a concern

Four kinds of concern, four doors. Pick the one that fits.

If someone is in danger, contact local emergency services first.

Your concernWhere to goStatusWhat happens next
Content or a user on CESWhen CES opens, use the report button on any post, profile or message. For urgent harm, [urgent route to be set by counsel].When CES opensUnder the Reporting and appeals policy (not yet published)
A decision about your contribution or rewardThe proposed Contributor Rights Charter sets out an appeal. Until it is adopted, write to governance@criscontinuity.tech.Charter approved for implementationUnder the Charter
An AI system, or harm it causedWrite to governance@criscontinuity.tech. Today the Board reads these messages. A dedicated AI incident channel is to be named.Read by the Board todayProposed Art. 21 path; response time to be confirmed
A security weaknessUse the contact in our security.txt file, under the vulnerability disclosure policy.Policy publishedWe aim to acknowledge within [three] working days and give a first assessment within [ten]
What is your concern about?Content or a user onCESReport button in the apponce CES opensUrgent: [counsel route]Policy: Reporting andappeals (not yet)A decision on my workAppeal under theContributor RightsCharter (proposed)Until adopted:governance@An AI system or harmgovernance@read by the Board todayLater: AI incidentchannel [name to come]Proposed Art. 16, 21A security weaknessSecurity contact insecurity.txtAcknowledge [3] daysAssess [10] daysSafe harbour: counselDanger to life: contact local emergency services first. The page says this above the four doors.
Raise a concern: four doors.

Governance

Arriving from an old governance link? The earlier governance pages have been replaced by this section.

CRIS is a Delaware public benefit corporation. Its public benefit is set out in the Third Article of its Certificate of Incorporation. The Board has 2 directors, both founders; there is no independent director yet. The accountability page sets out how independent Stewards would join. The Founding Charter is a separate document and is a draft; the bylaws are proposed. The AI Constitution and the Contributor Rights Charter are proposed, not yet adopted.

Our proposed bylaws would require every intelligent system to be tested against independent evidence before material deployment, and would bar any AI system from exercising corporate authority: see Article IX, AI Governance.

Oversight

Arriving from an old oversight link? The earlier oversight page described bodies that are not yet formed. This section shows what is in place and what is proposed.

Who can stop a CRIS system, and what happens when something goes wrong. Under our proposed AI Constitution, every CRIS system would have a named Responsible Steward who is accountable for it and can stop it at any time, and the Stop Authority could halt any AI system. Anyone could raise the alarm through the Article 21 channels. These are proposals, not yet adopted.

Board (two founders)In place · Delaware lawAudit CouncilproposedResponsible StewardproposedStop AuthorityproposedAssessorsnot engagedIncident path (proposed Art. 21)Anyone raises the alarmArt. 16.4; Art. 21 channelsSteward rates severity 1–4Schedule CSeverity 1: Board in [24 hours]Stop Authority may halt (16)Tell affected peopleand regulators where law asksWritten review in [30] daysLessons recordedSolid green: in place today. Dashed red: proposed or not engaged. Bracketed numbers are drafts.
Oversight: what is in place, what is proposed.
WhoStatus todayWhat it does
Board of StewardsIn office: the two foundersSets targets, writes the assessment, reads AI concerns today
Audit CouncilProposed; not formedProposed Art. 17.5; an interim independent reviewer named by the Board until it is appointed
Planetary and Africa Continuity CouncilsProposed; not formedPublish their own opinions
Responsible StewardProposed; not appointedProposed Art. 15.2: accountable for one system; receives incident reports
Stop AuthorityProposed; not appointedProposed Art. 16: may halt any CRIS system
Outside assessorsNot engagedAssess claims and standards

Status read from the public-benefit and trust data; set by a person with evidence.

Incidents. Under proposed Article 21, anyone who becomes aware of a possible incident would report it to the system’s Responsible Steward or the AI incident channel. Incidents would be rated by severity, from 1 (critical) to 4 (low), under Schedule C (draft). The most serious would reach the Board within [24 hours]; affected people would be told; a written review would follow within [30] days; and regulators would be notified where the law requires. Bracketed numbers stay bracketed until adopted. Until the Constitution is adopted, concerns about AI systems go to the Board, as above.

Read more: Human oversight (Part VI) and what a CRIS system will never do (Article 12).

Platform safety: CES and the Messenger

What our flagship products, CES Platform and CES Continuity Messenger, will do to protect people, and whether the policies behind it are published.

CES opens for its first pilot wave on 27 November 2026 (planned). When it opens, it will apply one standard everywhere: the strictest condition in every country where it operates. That is a design rule, not a claim of compliance.

  • Notice and action. When CES opens, anyone will be able to report a post, message, profile, group or advert, and reports will be handled under the Reporting and appeals policy.
  • Intimate images. Our target will be to act on a valid request within [24 hours], with identical copies found, as the Intimate-image and child-safety policy will set out.
  • Reasons and appeals. Every action will carry a statement of reasons and an appeal route.
  • Age. Age assurance at sign-up; adults only for paid pilot roles; no profiling-based adverts to minors.
  • Transparency. We plan to publish a transparency report twice a year, for each country, on reports, removals, appeals and legal requests; the first due date is to be set.
  • Records. Every moderation decision will be recorded with its reason.

These policies will be published here before CES opens:

CES policyStatus
Community standardsNot yet published
Reporting and appeals policyNot yet published
Intimate-image and child-safety policyNot yet published
Advertising policyNot yet published
Law-enforcement guidelinesNot yet published

5 of 5 are not yet published. If any is not published before Wave 1, what Wave 1 does without it is a decision for the Board and counsel.

[Counsel to confirm wording on intermediary liability, including Section 230 of the US Communications Act and each country’s conditions for platforms.]

12 NovToday16 NovDry run20 NovPolicies due27 NovWave 111 DecEast AfricaPolicyStatus on the pageCommunity standardsnot yet published (from data)Reporting and appeals policynot yet published (from data)Intimate-image and child-safety policynot yet published (from data)Advertising policynot yet published (from data)Law-enforcement guidelinesnot yet published (from data)Each row turns into a link when trust.json records the policy as published, with its date. Dates from CES data; 20 Nov is proposed.
CES: five policies before Wave 1.
Awaiting counsel. Community standards, reporting and appeals, the intimate-image and child-safety policy, the advertising policy and the law-enforcement guidelines are being finalised with US and country counsel, and will be published here before Wave 1.

Security

Where to report a weakness, and on what terms.

We publish a vulnerability disclosure policy and a security.txt file. We aim to acknowledge a report within [three] working days and give a first assessment within [ten]. Safe harbour for good-faith research is [subject to counsel].

Privacy

What we hold about you, and where it is described.

Our privacy page says who is responsible, what each CRIS service collects, and how to see, correct or delete it. None of our privacy notices is published yet; the privacy page is a draft for counsel. What CES will hold about you, and why, will be set out in the CES privacy notice in the CES legal centre; this page makes no claim about storage, encryption or regions beyond what those notices say.

Validation

Arriving from an old validation link? The earlier site showed validation badges that could not be checked. They have been removed; claims now appear here only with evidence.

Can you trust a claim on this site? Only if it names who checked it and links the evidence.

We do not call anything verified, certified or audited unless we name who did the work and link the evidence (proposed AI Constitution 23.2). No CRIS system has been independently assessed.

Every claim starts as Not assessed. It moves to Assessment underway only when an assessor is named under a written engagement, with scope and a start date, and becomes Assessed only with a linked report, a date and an expiry. The full registry page is planned for phase P3.

Not assessedDefault for every claimPage says so plainlyAssessment underwayAssessor named, scopeand start date linkedAssessedWho, when, scope, reportlinked; expiry date shownEach registry row holdsclaim · where it appears · status · assessor · scope · date · evidence link · expiresA word like verified, certified or audited may appear only when the row says Assessed.The build fails if the word appears anywhere without a matching Assessed row.Today: every row is Not assessed. The registry page ships in phase P3; the rule applies now.
The claims registry: a claim earns its word.

What this page does not yet show

A trust page earns trust by saying what is not yet true.

No independent directorTwo founders hold every Board seatNo independent assessmentNo system has been assessed by anyoneOversight bodies not formedAudit Council, Steward, Stop AuthorityNo incident log yetReporting not yet in placeCES policies not all publishedShown row by row, from dataLegal wording not signed offSection 230, safe harbour in bracketsEach limit disappears from the page only when its data record changes.
What this page does not yet show.
LimitToday
No independent directorBoth directors are founders. There is no independent director yet.
No independent assessmentNo CRIS system has been independently assessed.
Oversight bodies not formedThe Audit Council, the Responsible Steward and Stop Authority roles are proposed.
No incident logWe do not yet publish an incident log. An empty log would not be evidence of no incidents, so the page says reporting is not yet in place.
CES policiesThese CES policies are not yet published: Community standards; Reporting and appeals policy; Intimate-image and child-safety policy; Advertising policy; Law-enforcement guidelines.
Legal wordingSome legal wording on this page awaits counsel, and is shown in brackets.
Documents approved for implementationFounding Charter, Bylaws, AI Constitution, Contributor Rights Charter: approved by the Board for implementation on 20 November 2026; counsel is confirming the open points.

Generated from data; each row disappears only when its record changes. Where a date is set it is shown; elsewhere no date is promised.

Documents and history

Certificate of Incorporationfiled · Third Article: public benefitFounding CharterdraftBylawsproposedAI Constitutionproposed · Art. 16, 17, 21 and 23Contributor Rights CharterproposedVulnerability disclosure policypublished · terms bracketedGreen: in force or published. Red: draft or proposed. Each status is read from the document’s own data file.
Governing documents and their status.
DocumentStatus
Certificate of Incorporation (Third Article)Filed in Delaware, 22 December 2025
Founding CharterApproved by the Board for implementation, 20 November 2026
BylawsApproved by the Board for implementation, 20 November 2026
AI Constitution
Articles 16, 17, 21 and 23
Approved by the Board for implementation, 20 November 2026
Contributor Rights CharterApproved by the Board for implementation, 20 November 2026
Vulnerability disclosure policyPublished on the developer hub; response targets in brackets

This page replaced the earlier governance, oversight and validation pages in November 2026. The earlier site showed validation badges and status claims that could not be checked; they have been removed. Page owner: Governance and Compliance. Next review: 1 March 2027.

ItemStatusSourceBoard of directorsactivebenefit.json checksIndependent directornot yettrust.jsonAI Constitutionproposedai-constitution.jsonAudit Council and councilsproposedbenefit.json checksResponsible Steward, Stop Authorityproposedtrust.json rolesIndependent assessorsnot engagedbenefit.json checksCES policies (5)0 publishedtrust.json policiesVulnerability policypublished (terms in brackets)developer hubSystems statusall unsetstatus.json
Status today, read from data.

About this page

  1. Status from data. Every status word is read from data; no status is typed into the page.
  2. Proposed until adopted. The AI Constitution, the bylaws, the Contributor Rights Charter and every council are proposed until the Board adopts them and the adoption is recorded.
  3. Future tense before launch. CES commitments read “when CES opens” until the pilot is recorded as open.
  4. Name documents exactly. Certificate of Incorporation (Third Article); Founding Charter; article numbers for the AI Constitution.
  5. Say who. Each duty names the role that holds it; “we ensure” is not used.
  6. No unearned claims. Never verified, certified, audited or compliant without naming who and linking the evidence.
  7. Legal lines from counsel. Section 230, safe harbour, law-enforcement and regulator wording stays in brackets until counsel signs.
  8. Related organisations. Any mention of SCVX, LOGOS-C, BGTI or the Academy names it as a related organisation, also led by our Founder, with the related-party line; no partner is named before a written agreement.
RuleWriteNot1 Status from dataAudit Council: proposed (from data)Our Audit Council oversees2 Proposed until adoptedthe proposed AI Constitutionour AI Constitution requires3 Future tense before launchWhen CES opens, anyone will be able toAnyone can report; every report4 Name the document exactlyCertificate of Incorporation, Third Art.our public benefit charter5 Who, not “we ensure”The Responsible Steward can stop itWe ensure safety6 No unearned claimsNot yet independently assessedVerified, certified, audited7 Legal lines from counsel[counsel wording to come]We are protected by Section 2308 Related-party lineSCVX is a related organisation…our partner SCVX
Eight rules, with a do and a don’t.

Words to avoid

AvoidBecauseUse
safe, secure (as a state)A claim no one has assessedhow we report and fix problems
trusted byImplies endorsementName the agreement, or say nothing
24/7No rota existsThe response targets in brackets
AI-moderatedCES moderation design not finalreports are reviewed by people once the policy says so
guaranteeNo outcome promisesaim, target, with the number in brackets

Readers

CES user (Wave 1)How do I report harm,and what happens next?Answer: Raise a concern, CES rowContributorIf a decision about my workis wrong, where do I appeal?Answer: Raise a concern, rights rowSecurity researcherWhere do I send avulnerability, on what terms?Answer: Security sectionRegulator or counselWho is accountable, and whatis adopted versus proposed?Answer: Governance + statusFunder or partnerIs oversight real today,or a plan? Who checks it?Answer: Status today + limitsVisitor on an old linkWhere did the old badges go,and what replaced them?Answer: Arrival note per anchorEach box: who arrives, what they ask first, and where the page answers it.
Six readers, six first questions.
ReaderFirst questionAnswered in
CES userHow do I report harm, and what happens next?Raise a concern
ContributorWhere do I appeal a decision about my work?Raise a concern
Security researcherWhere do I send a vulnerability, under what terms?Security
Regulator or counselWho is accountable; what is adopted?Governance; documents
Funder or partnerIs oversight real today?Status; limits
Visitor on an old linkWhere did the old page go?An arrival note in each kept section

The road

1 Quick fixesby 24 Nov 2026Board line, tense, policystatus, security link, concerndoors, counsel brackets2 Rebuilt hubby 18 Dec 2026trust.json, status strip, eightsections, figures, arrivalnotes, restricted-word check3 Claims registryphase P3 (2027)Registry page from data,assessor rows, expiry, incidentlog if adoptedOrange: before Wave 1 (27 Nov). Blue: before the year ends. Green: with phase P3. Steps 1 and 2 are built on this preview.
The road: three steps.
StepByContentsToday
1 Quick fixes24 Nov 2026Board line; exact document names; when-CES-opens tense; policy status; counsel bracket on Section 230; security link; four concern doorsDone on this preview
2 Rebuilt hub18 Dec 2026trust.json; status strip and table; oversight and incidents; limits; arrival notes; figures; restricted-word checkDone on this preview
3 Claims registryPhase P3, 2027Registry page, assessor rows, expiry handling; incident log if Article 21 is adoptedOpen

How the page is built

Status strip: Board in place · bodies proposed · CES pilot dateFrom data1 Raise a concern: three doors + security#concern (new)2 Governance: who is accountable#governance (kept)3 Oversight: right to stop, incidents#oversight (kept)4 Platform safety: CES, when it opens#platform-safety (kept)5 Security and privacy#security, #privacy (new)6 Validation: the claims registry#validation (kept)7 What this page does not yet show#limits (new)8 Documents and history#documents (new)
The rebuilt hub, top to bottom.
SectionAnchorKept or new
Raise a concern#concernNew
Governance#governanceKept (3 redirects)
Oversight#oversightKept (3 redirects)
Platform safety#platform-safetyKept
Security#securityNew
Privacy#privacyNew
Validation#validationKept (3 redirects)
Limits#limitsNew
Documents and history#documentsNew
/governance, /governance/archive,/governance/*#governancearrival note/compliance/oversight, /oversight,/oversight/*#oversightarrival note/validation, /validation/archive,/validation/*#validationarrival note16 site links (to the top and keptanchors)top, #governance, …unchanged
Old routes and links land on kept anchors.
data/trust.json (new)data/benefit.jsondata/ces-platform.jsondata/status.jsonBuildtrust.md + data/trust/ pagestrip, table, doors, limitsChecks (fail the build)words, anchors, tense, datesDocument status is read from each document’s own data file. No status word is typed in trust.md; the build fails if one is.
Where every status comes from.
Status strip401 Raise a concern2002 Governance1203 Oversight2004 Platform safety2005 Security and privacy1206 Validation1207 Limits + 8 Documents100 + tablesAbout 1,300 words in all; each section opens with a one-sentence answer.
Word budget, about 1,300 words.
Authordraft + checksGovernanceand ComplianceCounsellegal linesBoardstatus wordsPublishbuild passesCounsel signs: Section 230, safe harbour, the urgent CES route, law-enforcement wording.The Board confirms: every status word, the Board composition line, the limits list.Quick fixes need Governance and counsel only; the rebuilt hub needs all four.
Approvals before publishing.
criscontinuity.tech/trust/Trust and safetyHow we govern our systems, who oversees them, and how to raise a concern.In place: the Board and the law · Proposed: see the status table · CES pilot: 27 Nov 2026Raise a concernIf someone is in danger, contact local emergency services first.On CESReport in the appwhen CES opensHow it worksMy contributionCharter appeal(proposed)Read the draftAn AI systemgovernance@read by the BoardWhat happens nextSecuritysecurity.txtreply in [3] daysDisclosure policyMock-up only. Counts and dates are drawn from data at build time.
Mock-up of the top of the rebuilt page.
CheckFails when
ts-words“verified”, “certified”, “audited”, “independently assessed” or “compliant” appear without an Assessed claim row
ts-typedA status word is typed in trust.md outside a data directive
ts-tense“anyone can report” or present-tense CES commitments appear before the pilot is recorded open
ts-anchorsAny of the four kept anchors is missing
ts-redirectsA redirect targets a missing anchor on this page
ts-bracketsA counsel item is pending but its text is not bracketed
ts-relatedSCVX, LOGOS-C, BGTI or the Academy appears without the related-party line
ts-datesA date on the page differs from its data file

All eight run in every build.

What changed

Governanceboard, charter; bylaws draftBoard is two founders, not said (K1); nine oldroutes land here (K10)OversightStewards, Stop Authority (proposed)Keep; add incidents, Art. 21 (K11)Platform safety: CESsix commitments; Section 230; policies before Wave 1Present tense before launch (K3); policies due,status unknown (K2)Validationno unearned claims; registry in P3Keep the rule; cite its source; registry from data(K6)Missing: security and security.txt (K7) · three concern routes (K8) · status today (K9) · privacy linkGreen: keep. Red: fix.
The page before this update: four sections, about 400 words.
#FindingFixed by
K1Board composition not saidGovernance: two directors, both founders
K2CES policies due, status unknownPolicy rows from data; limits
K3Present tense before launch“When CES opens”, switched by data
K4Section 230 paragraph without counselBracketed for counsel
K5“Public benefit charter”Certificate of Incorporation (Third Article); Founding Charter
K6Validation rule unsourcedProposed Art. 23.2 named; claims states; build check
K7No securitySecurity section; security.txt
K8One route for concernsFour doors
K9No statusStatus strip and documents table from data
K10Nine old routesArrival notes on kept anchors
K11Incidents missingOversight table and the Article 21 path
K12No figuresSeventeen figures

Owner: Governance and ComplianceNext review: 2027-03-01