Compliance & AI Safety

Regulatory Framework

The Institute for Cognitive Sovereignty operates under the governance of Holistic Quality LLC. We align our practices with applicable regulations and proactively adopt emerging AI governance standards. The Institute is an advocacy and public-education organization — a publisher of investigations that cites the Research of Record — and an AI deployer: we use commercial AI models as tools and train or fine-tune no models of our own for the published corpus.

Framework posture last reviewed 2026-06-05. "Aligned" denotes voluntary, self-assessed alignment, not certification.

Framework	Scope & our posture	Status (as of 2026-06-05)
GDPR Principles	Data protection, privacy by design	Active
CCPA Alignment	California consumer privacy	Active
EU AI Act (Reg. 2024/1689)	As a deployer with transparency duties — not a general-purpose-AI (GPAI) provider. GPAI provider obligations (Art. 53) bind those who train and place models on the market and applied from August 2025; the high-risk regime and most remaining provisions apply from August 2026. We train no models, so the provider duties do not attach to us.	Monitoring & preparing
NIST AI RMF 1.0	AI risk management framework (GOVERN / MAP / MEASURE / MANAGE)	Aligned (self-assessed)
NIST AI 600-1 — Generative AI Profile (Jul 2024, Mar 2025 updates)	GenAI-specific risk overlay on the RMF; our human-anchor review and claim-screening stack map most strongly to the MEASURE and MANAGE functions	Aligned (self-assessed)
Emerging US AI Legislation	Federal/state AI requirements	Monitoring

Upstream model providers publish their own frontier-safety frameworks; our AI-assisted work inherits those provider-side safeguards in addition to our own human-anchor review.

AI Safety Principles

Our investigations extensively use AI systems (the Hexad methodology). We hold ourselves to strict safety standards:

• 1 Transparency: We disclose AI involvement in our work. The Hexad methodology is documented. AI contributions are acknowledged, not hidden.
• 2 Screening: AI-generated claims are automatically screened and their citations checked against primary sources — a fabrication-prone pre-filter, pending human verification, not verification itself.
• 3 Limitation Awareness: We acknowledge what AI can't do—make ethical judgments, verify its own outputs, or replace domain expertise.
• 4 No Autonomous Publishing: No AI output is published without human review. Period.
• 5 Beneficial Use: We use AI to help people, not exploit them. Our investigations aim to protect cognitive sovereignty, not undermine it.

Hexad Methodology Safeguards

The Hexad methodology (human + 5 AI systems) includes built-in safeguards:

Intellectual Property & AI Training

We take a clear stance on AI training data:

Our Content

• Our investigations and other content may not be used to train AI models without explicit license
• This includes scraping for training datasets
• We consider unauthorized AI training on our work to be copyright infringement

Our AI Use

• We use commercially licensed AI services (Claude, GPT, Gemini, etc.)
• We don't train our own models on others' work without permission
• We respect robots.txt and terms of service

Incident Response

If we discover that our content contains errors, harmful information, or violations of these commitments:

• Immediate: Remove or correct the content
• Transparent: Acknowledge the error publicly
• Root cause: Identify how the failure occurred
• Preventive: Update processes to prevent recurrence

Report concerns: compliance@cognitivesovereignty.institute

Future Regulatory Readiness

AI regulation is evolving rapidly. We're preparing for:

• Disclosure requirements for AI-generated content
• Watermarking and provenance standards
• Risk assessment requirements
• Transparency obligations

Our philosophy: Don't wait for regulations. Do the right thing now. When rules arrive, we'll already be compliant.