Ethical AI Manifesto
Our contractual and moral commitment for AI in service of humanity. 6 non-negotiable principles inspired by Constitutional AI: human first, radical transparency, data protection, safety, long-term vision, accountability.
Aggil Manifesto for Ethical Artificial Intelligence
📜 Founding Document
Our non-negotiable commitments to develop AI that serves humanity long-term, inspired by Constitutional AI principles.
Our Fundamental Conviction
"Artificial intelligence represents one of the most transformative technologies of our time. Its value is not measured solely in productivity gains, but in sustainable positive contribution to society."
At Aggil, we have been developing AI systems for over 17 years. Over these years, we have observed that the most performant AI is one that respects humanity, protects privacy and remains transparent.
This manifesto formalizes our action principles, inspired by Constitutional AI research and international best practices in AI ethics.
Our 6 Fundamental Commitments
Human First, Always
AI systems that augment human capabilities without replacing them. Final human validation on critical decisions. Automation frees time for high-value tasks.
Radical Transparency
No "black box". Complete technical documentation provided. Natural language explanations of decisions. Auditability guaranteed at all times.
Absolute Data Protection
Systematic privacy-by-design. Data minimization and anonymization. AES-256 encryption standard. Sovereign hosting available.
Uncompromising Security
Rigorous adversarial testing. Continuous 24/7 monitoring. Fail-safes on critical scenarios. Documented contingency plans.
Positive Long-Term Impact
5-10 year vision, not quick wins. Evolutionary and maintainable systems. Societal and environmental impact measurement.
Assumed Responsibility
Clear and contractual accountability. AI governance provided. Non-ethical project refusal. Guaranteed responsive support.
1. Human First, Always
We develop AI systems that augment human capabilities without replacing them.
Systematic Human-in-the-Loop
ArchitectureWorkflows integrating configurable human validation points
Training & Empowerment
SupportComprehensive training for complete system mastery
Destructive Project Refusal
EthicsMissions aiming at job elimination without retraining refused
Progressive Control
GovernanceControl takeover possible at any time by your teams
2. Radical Transparency
You understand how your AI systems work and make their decisions.
We provide:
Technical Documentation
Complete architecture, algorithms, documented parameters
Explainability Dashboards
Real-time visualization of AI reasoning
Audit Logs
Complete logging queryable at will
Training Sessions
Regular reviews to guarantee team understanding
3. Absolute Data Protection
Your data is your most precious asset. We protect it like our own.
Architecture
Privacy-by-Design
Design integrating protection from origin
Compliance
Data Minimization
Collection only of strictly necessary data
Right to be Forgotten
Complete GDPR Article 17 implementation
Security
AES-256 Encryption
Military-grade encryption standard on all data
Infrastructure
Sovereign Hosting
France/EU options for strict compliance
Protection
Anonymization
Advanced anonymization techniques applied
4. Uncompromising Security
Performant AI is reliable and secure AI.
Testing Phase
Unit, integration, adversarial tests, red-teaming
Validation Phase
Security, performance, compliance validation
Monitoring Phase
24/7 surveillance, automatic alerts, real-time dashboards
Improvement Phase
Continuous optimization, security patches, evolutions
Our security guarantees:
5. Positive Long-Term Impact
We build to last, not to impress during a demo.
Impact Analysis
Societal, environmental, economic impact assessment before project
Evolutionary Architecture
Modular systems easily maintainable and extensible
3-5 Year Roadmap
Long-term vision with clear and measurable milestones
Continuous Measurement
Positive impact KPIs tracked and optimized continuously
""We systematically refuse projects whose main objective is job elimination without retraining plan, or whose societal impact would be negative.""
6. Assumed Responsibility
We assume full and complete responsibility for our systems.
AI Governance
GovernanceComplete framework provided: processes, roles, responsibilities
Project Ethics Committee
OrganizationEstablishment of multidisciplinary committee on critical projects
Contractual Documentation
ContractsClear accountability: who does what, who decides what
Responsive Support
SupportSupport SLA: <2h critical issues, <24h others
Exit Clause
ProtectionPossibility to stop project if ethical drifts observed
Our Contractual Commitment
This manifesto is not just a marketing document. It is integrated into our client contracts.
| Commitment | Contractual Clause |
|---|---|
| Human-in-the-Loop | Human validation points documented and implemented |
| Transparency | Complete documentation access + explainability dashboards |
| Data Protection | Certified GDPR compliance + regular audits |
| Security | 99.9% uptime SLA + contingency plan provided |
| Positive Impact | Societal impact analysis included in each project |
| Responsibility | Clear accountability + ethical exit clause |
Questions About Our Manifesto
Is this manifesto legally binding?
Yes. These principles are contractually integrated into all our client engagements. If we don't respect our commitments (e.g., transparency default, security breach through negligence), you have contractual recourse including exit clauses.
What happens if a project drifts from these principles?
Our contracts include an ethical exit clause: if we or you observe a drift from these principles (e.g., discriminatory bias detected, non-ethical use identified), the project can be suspended or stopped without penalties. Protection of our values takes priority over revenue.
How to verify that you really respect these principles?
We provide for each project: (1) complete technical documentation, (2) access to explainability dashboards, (3) queryable audit logs, (4) quarterly compliance reports, (5) annual security audits by independent third party. You have total oversight of our systems.
Why refuse profitable projects?
We have refused several lucrative projects in recent years because they didn't respect our principles (e.g., invasive employee surveillance, job elimination without retraining). Our reputation and values are worth more than short-term revenue. Responsible AI is our differentiator, not a marketing slogan.
Do You Share These Values? Let's Work Together
If this manifesto resonates with your AI vision, let's discuss how we can collaborate on your projects.
📜 Manifesto published: January 2025 • ✅ Contractually binding • 🔄 Updated annually