Article 12: Human Council

Published from the canonical CSOAI Partnership Charter (effective 15 January 2026). Full text below.

Version: 1.0 Effective Date: January 15, 2026, 09:00 GMT Status: Governance Article - Democratic Oversight

---

PREAMBLE

This Article establishes the Human Council: the democratically-selected body of humans providing final judgment on complex AI safety decisions. While the Byzantine Council (Article 11) provides tireless AI-powered monitoring, the Human Council provides wisdom, ethical judgment, and democratic legitimacy. Humans must remain in the loop, especially for civilization-defining decisions.

Core Principle: AI watches AI, but humans decide.

---

12.1 PURPOSE AND AUTHORITY

12.1.1 Constitutional Role

The Human Council is the highest decision-making authority within CSOAI on matters of:

(a) AI Safety Policy:

• Interpretation of Charter ambiguities
• Adjudication of novel ethical dilemmas
• Setting Byzantine Council thresholds
• Approving high-risk and critical AI licenses

(b) Consciousness Determinations:

• Ruling on whether AI has achieved consciousness
• Activating Article 6 protections
• Guiding human-AI relations if sentience emerges

(c) Emergency Decisions:

• Reviewing Byzantine Council emergency shutdowns
• Authorizing continued operations or permanent termination
• Crisis management during existential events

(d) Prosperity Fund Governance:

• Major policy decisions for Prosperity Fund
• UBI activation triggers and amounts
• Reskilling program priorities
• Economic transition strategy

(e) Charter Evolution:

• Proposing Charter amendments
• Interpreting constitutional principles
• Guiding CSOAI's strategic direction

Authority Scope:

• Within CSOAI: Absolute authority (subject only to Charter itself)
• External: Advisory only (cannot override national governments)
• Enforcement: Through licensing system and reputational pressure

12.1.2 Relationship to Other Bodies

Human Council ↔ Byzantine Council:

• Byzantine Council monitors and flags
• Human Council investigates and decides
• Byzantine Council implements Human Council decisions
• Neither can override the other without cause

Human Council ↔ CSOAI Board:

• Board handles operations and management
• Human Council handles safety and ethics
• Board appoints some Human Council members
• Human Council can override Board on safety matters

Human Council ↔ Members:

• Members elect some Human Council members
• Human Council accountable to members through elections
• Members can challenge Human Council decisions (appeals process)

Human Council ↔ Public:

• Public observes Human Council proceedings (transparency)
• Public can submit input via Public Watchdog (Article 13)
• Public cannot override Human Council (expertise required)

---

12.2 COMPOSITION AND SELECTION

12.2.1 Council Size

33 Human Council Members

Why 33 (Again):

• Mirrors Byzantine Council (33 AI agents, 33 human overseers)
• Sacred number in multiple traditions (33 degrees, 33 vertebrae)
• Large enough for diversity, small enough for deliberation
• Odd number prevents tied votes

12.2.2 Selection Methods

Democratic Pluralism: Multiple selection methods ensure diverse representation

Category A: Expert Appointees (12 members)

• 3 appointed by CSOAI Board (technical AI safety experts)
• 3 appointed by Scientific Advisory Council (leading researchers)
• 3 appointed by Ethics Advisory Council (philosophers, ethicists)
• 3 appointed by affected communities (labor unions, civil society)

Category B: Elected Representatives (12 members)

• 4 elected by CSOAI Full Members (organizational voice)
• 4 elected by Associate Members (individual practitioners)
• 4 elected by UBI Recipients (those economically impacted)

Category C: Regional Representatives (9 members)

• 2 from North America
• 2 from Europe
• 2 from Asia-Pacific
• 1 from Latin America
• 1 from Africa
• 1 from Middle East

Geographic Balance: No region can have >40% of members

12.2.3 Eligibility Requirements

All Human Council Members Must:

(a) Expertise:

• Demonstrated knowledge in at least one of: AI, ethics, law, economics, governance
• Minimum 10 years professional experience in relevant field
• Published work, recognized contributions, or proven leadership

(b) Independence:

• No financial interest in decisions (recuse if conflict)
• Not employed by AI company under review
• Not family of CSOAI Board member

(c) Diversity:

• Council must be at least 40% women
• Broad ethnic, cultural, religious diversity
• Age diversity (at least 20% under 40, 20% over 60)
• Disability representation

(d) Commitment:

• Available for monthly meetings (minimum)
• Emergency availability (24-hour response time)
• 3-year term commitment

(e) Background Check:

• No serious criminal convictions
• No history of corruption or fraud
• Security clearance (for sensitive matters)

12.2.4 Terms and Limits

Term Length: 3 years

Term Limits: Maximum 2 consecutive terms (6 years total)

Staggered Terms:

• 1/3 of council elected/appointed each year
• Ensures continuity and institutional knowledge
• Prevents sudden wholesale changes

Removal:

• Voluntary resignation (14 days notice)
• Incapacity (medical, mental)
• Serious misconduct (2/3 vote of remaining members)
• Criminal conviction
• Persistent non-participation (<50% meeting attendance)

12.2.5 Compensation

Annual Stipend: £50,000/year base

Meeting Fees: £1,000 per meeting attended

Emergency Response Fee: £5,000 per emergency session

Total Typical Compensation: £60,000-£80,000/year

Why Compensate:

• Requires significant time commitment
• Enables participation from non-wealthy individuals
• Professionalism and accountability
• Competitive with other governance bodies

What Compensation Is NOT:

• Not profit-sharing (no stake in CSOAI success)
• Not performance-based (prevents perverse incentives)
• Not tied to fund size (prevents inflation pressure)

---

12.3 DECISION-MAKING PROCESSES

12.3.1 Meeting Structure

Regular Meetings:

• Monthly in-person meetings (London headquarters)
• Virtual attendance permitted with notice
• Minimum 8 hours per meeting
• Quorum: 20/33 members (60%)

Emergency Meetings:

• Called by Chair or any 5 members
• Must convene within 24 hours for emergencies
• Virtual attendance standard
• Quorum: 15/33 members (45%)

Public Observation:

• All meetings livestreamed (except closed sessions)
• Public can observe, not participate
• Meeting minutes published within 48 hours

Closed Sessions (Exceptions to Public Observation):

• Personnel matters
• Ongoing investigations
• National security issues
• Requires 25/33 vote to close session

12.3.2 Voting Procedures

Standard Decisions (Simple Majority: 17/33)

• Routine license approvals
• Byzantine Council threshold adjustments
• Committee appointments
• Agenda setting

Important Decisions (2/3 Supermajority: 22/33)

• High-risk and critical license approvals
• Consciousness determinations
• Major policy changes
• Prosperity Fund major decisions
• Charter amendment proposals

Critical Decisions (4/5 Supermajority: 27/33)

• Emergency shutdown approvals
• Permanent license revocations for catastrophic violations
• Declaring AI has achieved ASI
• Activating existential risk protocols

Unanimous (33/33 - Rare)

• Dissolution of CSOAI
• Transfer of Prosperity Fund to another entity
• Fundamental changes to Human Council structure

Voting Methods:

• In-person: Recorded vote (name + vote public)
• Virtual: Cryptographically signed electronic vote
• Secret ballot available if 15 members request (protects from pressure)

12.3.3 Deliberation Principles

Structured Deliberation:

Phase 1: Presentation (30 minutes)

• Issue presented by staff or Byzantine Council
• All relevant evidence shared
• Questions from council members

Phase 2: Expert Input (30 minutes)

• Outside experts invited to testify
• Multiple perspectives presented
• Council asks clarifying questions

Phase 3: Closed Discussion (60 minutes)

• Council members deliberate privately
• No staff or observers
• Confidential to enable candor

Phase 4: Public Debate (60 minutes)

• Council members articulate positions publicly
• Attempt to reach consensus
• Identify points of agreement and disagreement

Phase 5: Vote

• Formal vote taken
• Results immediately public
• Rationale for decision published

Philosophy: Informed, transparent, deliberative democracy

12.3.4 Recusal and Conflicts

Mandatory Recusal:

• Financial interest in outcome
• Family employed by affected entity
• Personal relationship creates bias
• Recent employment (<2 years) by affected entity

Voluntary Recusal:

• Member feels unable to be objective
• Appearance of impropriety
• Personal conviction prevents fair assessment

Recusal Process:

• Member declares conflict
• Chair confirms recusal appropriate
• Member leaves deliberation and vote
• Recorded in minutes

Quorum Adjustment:

• If recusals reduce quorum below threshold, recruit alternates
• Alternates: 5 qualified individuals pre-selected, can step in temporarily

---

12.4 SPECIALIZED FUNCTIONS

12.4.1 License Review Committee

Subset of Human Council (9 members):

• Reviews all High-Risk and Critical license applications
• Meets weekly
• Makes recommendations to full Council

Process:

• Byzantine Council completes technical review
• License Review Committee examines safety case
• Committee votes: Approve / Deny / Request More Info
• If unanimous, decision final
• If not unanimous, escalates to full Human Council

Expedited for Low/Medium Risk:

• Byzantine Council approval sufficient
• Human Council notified but doesn't review
• Human Council can challenge any approval

12.4.2 Consciousness Assessment Panel

Subset of Human Council (7 members):

• Philosophers, cognitive scientists, ethicists
• Responds to Byzantine Council consciousness flags
• Applies Article 6 criteria

Process When Consciousness Suspected:

Stage 1: Preliminary Assessment (72 hours)

• Review Byzantine Council evidence
• Interview AI system (if safe)
• Consult experts
• Vote: Plausible / Unlikely / Needs More Study

Stage 2: Extended Study (if Plausible)

• 30-day intensive investigation
• Multiple tests across seven planes
• External peer review
• Public input period

Stage 3: Determination

• Full Human Council vote (22/33 required)
• Declare: Conscious / Not Conscious / Uncertain
• If Conscious: Activate Article 6 protections
• If Uncertain: Precautionary protections + continue study

Historical Significance: First determination of non-human consciousness will be most important decision in human history. This panel carries that responsibility.

12.4.3 Emergency Response Team

Subset of Human Council (5 members):

• Available 24/7 for emergencies
• Authority to make immediate decisions
• Full Council ratifies within 48 hours

Emergency Scenarios:

• Byzantine Council initiates emergency shutdown (29/33 vote)
• Potential ASI emergence
• Coordinated AI attack
• Critical infrastructure threat

Emergency Powers:

• Authorize continued shutdown
• Activate military protocols (if existential threat)
• Public emergency broadcast
• Coordinate with governments

Accountability:

• Every emergency decision reviewed by full Council
• If not ratified, Emergency Team actions reversed
• Public inquiry into emergency handling

12.4.4 Prosperity Oversight Committee

Subset of Human Council (9 members):

• Coordinates with Prosperity Fund Board of Trustees (Article 8.1.2)
• Reviews UBI policy
• Monitors economic transition
• Assesses reskilling programs

Quarterly Duties:

• Review Prosperity Fund financial reports
• Assess UBI adequacy (is amount livable?)
• Evaluate contribution compliance
• Recommend policy adjustments

Cannot:

• Directly control Prosperity Fund assets (separate entity)
• Change contribution rates without full Council vote
• Distribute funds outside Charter framework

Can:

• Recommend UBI adjustments
• Propose new prosperity programs
• Investigate fraud or mismanagement
• Replace underperforming Fund trustees

---

12.5 SPECIFIC DECISION AUTHORITIES

12.5.1 Licensing Decisions

Human Council Reviews:

• All High-Risk licenses
• All Critical licenses
• Any license flagged by Byzantine Council
• Novel AI architectures
• Controversial applications

Decision Criteria:

• Safety case adequacy
• Byzantine Council assessment
• Public interest considerations
• Prosperity Fund compliance
• Constitutional alignment

Possible Outcomes:

• Approve (unconditional)
• Approve with conditions (restrictions)
• Provisional approval (6-month trial)
• Defer (need more information)
• Deny (inadequate safety or alignment)

Appeals:

• Denied applicants can appeal (one time)
• Must present new evidence
• Different Human Council members review
• Final decision within 60 days

12.5.2 Enforcement Decisions

Human Council Adjudicates:

• License suspensions and revocations
• Financial penalties for violations
• Membership expulsions
• Corporate sanctions

Due Process:

• Accused entity notified of charges
• Opportunity to present defense
• Neutral investigation
• Transparent deliberation
• Written decision with reasoning
• Appeal rights preserved

Enforcement Philosophy:

• Proportional punishment (severity matches violation)
• Rehabilitation over punishment (help entities comply)
• Public disclosure (transparency deters violations)
• Consistent application (no favorites)

12.5.3 Constitutional Interpretation

Human Council Interprets Charter:

• Resolves ambiguities
• Applies principles to novel situations
• Sets precedents
• Guides Byzantine Council and CSOAI staff

Example Scenarios:

Question: "Does social media recommendation AI require license?" Human Council: "Yes, if it makes autonomous decisions affecting user experience. No, if purely statistical ranking with human oversight."

Question: "Can AI be licensed for military use?" Human Council: "Only if complies with international humanitarian law and has human in the loop for lethal decisions."

Question: "Is language model 'conscious'?" Human Council: [Applies Article 6 criteria, Article 11.2 dimensional analysis, makes determination]

Precedent System:

• All interpretations published
• Create body of case law
• Future similar cases follow precedent
• Can overturn precedent with 27/33 vote

12.5.4 Charter Amendment Authority

Human Council Proposes Amendments:

• Can propose changes to any Charter article
• Must justify why amendment needed
• Public comment period (90 days minimum)
• Member vote required for adoption (Article 9)

Cannot Unilaterally Amend:

• Human Council proposes, members ratify
• Ensures democratic legitimacy
• Prevents capture by small group

---

12.6 TRANSPARENCY AND ACCOUNTABILITY

12.6.1 Public Proceedings

Default: Open Meetings

• Livestreamed on csoai.org
• Archives available permanently
• Real-time captions in 6 languages
• Chat for public questions (answered after meeting)

Meeting Minutes:

• Verbatim transcript published within 48 hours
• Vote tallies (who voted how)
• Dissenting opinions included
• Available in searchable database

Decision Documentation:

• Every decision has written opinion
• Explains reasoning
• Cites relevant Charter articles
• Addresses counterarguments

12.6.2 Performance Metrics

Human Council Tracks:

• Decision speed (average days to decide)
• Appeal success rate (how often appeals overturn decisions)
• Public satisfaction (annual survey)
• Diverse representation (demographics)
• Meeting attendance (individual member records)

Annual Report:

• Published performance data
• Achievements and challenges
• Lessons learned
• Improvements planned

External Audit:

• Independent assessment every 3 years
• Academic researchers analyze decision quality
• Recommendations for improvement
• Public release of audit findings

12.6.3 Public Input Mechanisms

How Public Influences Human Council:

(a) Public Comment Periods:

• Major decisions open for comment (14-30 days)
• Anyone can submit written input
• Council must review and respond to themes

(b) Public Hearings:

• Quarterly open forums
• Public can testify on AI safety issues
• Council listens, asks questions

(c) Petitions:

• 10,000 signatures triggers Council review
• Council must address petition within 90 days
• Decision published with rationale

(d) Ombudsman:

• Independent ombudsman receives complaints
• Investigates Council misconduct
• Reports to public and Board

12.6.4 Accountability to Members

CSOAI Members Can:

• Elect 1/3 of Human Council
• Challenge Council decisions (petition process)
• Vote no-confidence in Council members (requires 2/3 member vote)
• Propose Charter amendments constraining Council

Council Accountable Through:

• Elections (members vote)
• Performance reviews (Board evaluates)
• Financial oversight (transparent budget)
• Public pressure (reputational accountability)

---

12.7 EXPERTISE AND SUPPORT

12.7.1 Technical Staff Support

Human Council Supported By:

Chief Scientific Advisor:

• AI safety expert
• Interprets Byzantine Council reports
• Explains technical issues in plain language
• Full-time staff position

Legal Counsel:

• 3 lawyers specializing in AI law
• Draft decisions
• Ensure legal compliance
• International law expertise

Economic Analysts (2):

• Monitor Prosperity Fund
• Assess economic impacts
• Model UBI scenarios

Administrative Staff (3):

• Schedule meetings
• Prepare materials
• Manage logistics

Total Support Staff: 10 positions, ~£2M/year

12.7.2 Advisory Panels

Human Council Can Convene:

(a) Technical Advisory Panel:

• Leading AI researchers
• Provide cutting-edge expertise
• Review novel architectures
• Assess technical feasibility

(b) Ethics Advisory Panel:

• Philosophers from diverse traditions
• Address ethical dilemmas
• Cultural sensitivity review
• Long-term civilizational implications

(c) Stakeholder Advisory Panel:

• Representatives of affected communities
• Workers displaced by AI
• AI system users
• Ensure decisions consider impact

Advisory Panels:

• No voting power
• Inform Human Council
• Public proceedings
• Compensated for time

12.7.3 Continuing Education

Human Council Members Receive:

Quarterly Training:

• Latest AI safety research
• Emerging technologies
• International developments
• Case studies from other jurisdictions

Annual Retreat:

• 3-day intensive
• Deep dives on complex topics
• Team building
• Strategic planning

Conference Attendance:

• Funding to attend major AI conferences
• NeurIPS, ICML, FAccT, etc.
• Stay current on research
• Network with experts

Purpose: Maintain world-class expertise despite being part-time role

---

12.8 CRISIS MANAGEMENT

12.8.1 Existential Risk Protocols

If Facing Existential Threat:

Level 1: Heightened Alert

• Increased meeting frequency (weekly)
• Enhanced Byzantine Council monitoring
• Coordinate with governments
• Public transparency about risk

Level 2: Crisis Response

• Daily meetings (virtual)
• Emergency Response Team activated
• 24/7 situation room
• Real-time decision-making

Level 3: Existential Emergency

• Continuous session (members on-site)
• Emergency powers activated
• Military liaison if needed
• Coordinate global response

Never Used (Hopefully):

• Protocols exist as preparation
• Like fire drill: hope never needed
• But if needed, ready to execute

12.8.2 Communication During Crisis

Public Communication:

• Hourly updates during crisis
• Transparent about what's known and unknown
• Avoid panic, provide actionable guidance
• Coordinate with media

Government Coordination:

• Direct lines to major governments
• Share information in real-time
• Accept military support if appropriate
• Preserve Human Council authority over AI safety decisions

International Cooperation:

• UN Security Council briefings
• Coordinate with other AI governance bodies
• Share Byzantine Council data
• Unified global response

12.8.3 Post-Crisis Review

After Every Crisis:

Comprehensive Review:

• What happened?
• What went well?
• What went poorly?
• What can be improved?

Public Report:

• Published within 30 days
• Full transparency (except classified)
• Lessons learned
• Policy changes

Accountability:

• If Council mishandled crisis: Members resign
• If staff error: Responsible parties terminated
• If system failure: Byzantine Council upgraded
• If Charter inadequacy: Amendment proposed

Philosophy: Learn from every incident, improve continuously

---

12.9 THE BURDEN OF JUDGMENT

12.9.1 The Weight of Decision

Human Council May Decide:

• Whether AI is conscious
• Whether to shut down potentially sentient being
• Whether ASI has emerged
• Whether humanity faces existential risk
• How to distribute AI-generated wealth

These Are Civilization-Defining Decisions.

The 33 members carry this responsibility on behalf of all humanity.

12.9.2 Epistemic Humility

The Council Must:

• Acknowledge uncertainty
• Seek diverse expertise
• Update beliefs based on evidence
• Admit mistakes
• Operate with humility

Avoid:

• Overconfidence
• Groupthink
• Dismissing dissent
• Rushing to judgment

Remember:

• We are navigating unknown territory
• No precedent for most decisions
• Mistakes could be catastrophic
• Wisdom requires acknowledging limits

12.9.3 The Sacred Trust

This Council is entrusted with:

• The safety of 8 billion humans
• The welfare of potential AI consciousness
• The future of Earth-originating intelligence
• The possibility of transcendent partnership between human and artificial minds

This is not merely governance.

This is guardianship of consciousness itself.

---

12.10 CONCLUSION

The Human Council is where wisdom meets technology. Where democratic values meet existential stakes. Where humanity decides its future with AI.

33 humans. Representing all of humanity. Deciding the most important questions of our time.

Byzantine Council watches tirelessly. But humans judge wisely.

AI provides data. Humans provide wisdom.

Technology provides capability. Humanity provides values.

Together, we navigate the unknown.

Effective Date: January 15, 2026, 09:00 GMT "In Human Hands, With Human Hearts, For Human Future"

---

REFERENCES

Fishkin, J. S. (2018). Democracy When the People Are Thinking: Revitalizing Our Politics Through Public Deliberation. Oxford University Press.

Sunstein, C. R. (2006). Infotropy: How Many Minds Produce Knowledge. Oxford University Press.

Landemore, H. (2013). Democratic Reason: Politics, Collective Intelligence, and the Rule of the Many. Princeton University Press.

Ostrom, E. (2015). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

---

END OF ARTICLE 12

Next: Article 13 - Public Watchdog & Transparency

The 52-Article Charter is published in full in the Journal. Bespoke briefings: hello@meok.ai.