AI SAFE² secures the AI system. But who’s securing the human operating it?
That’s the question the Cognitive Sovereignty Framework answers. v2.0 is the full doctrine release: a model of human cognitive resilience, a taxonomy of what attacks it, and a protocol stack for defending it. If AI SAFE² is about making agents trustworthy, CSF is about keeping people sovereign in a world of increasingly persuasive machines.
What’s new in v2.0
Six-domain resilience model. Physical, psychological, social, and spiritual fitness, plus the new Domain 6: Digital & AI Symbiosis. The relationship between humans and AI systems isn’t a side issue anymore; it’s a domain of resilience in its own right, and it needs training like any other.
The Cognitive Threat Taxonomy, 41 entries across 5 causal layers. From civilizational drivers down through substrate threats, active manipulation techniques, and delivery mechanisms, to human outcomes. Each entry gets quantitative severity scoring via CTSS, with a calculator included. Cognitive threats finally get the same structured treatment as software vulnerabilities.
EFA and the E7 protocol stack. Ethical Functionality without Agency, AI systems that serve their function without developing or exercising their own agency, implemented through the E7 stack, with Mission and Authority permanently anchored at the human Layer 7. The 8-step operational sequence makes it deployable, not just philosophical.
Honestly documented edge cases. Full infrastructure capture, autonomous swarm reality formation, civilizational reproduction failure. v2.0 names the scenarios where the framework strains, instead of pretending they don’t exist. A doctrine that admits its limits is one you can trust.
Dual-licensed MIT + CC-BY-SA 4.0, because defensive knowledge should spread.
“AI SAFE² secures the AI system; CSF defends the human operating it. Neither framework is sufficient without the other.”
Why this matters
- Cognition is the primary attack surface of the 21st century. Models get the security budget; minds get nothing. v2.0 corrects that.
- Voluntary cognitive atrophy is the quiet threat. The framework argues that willingly outsourcing our thinking is more dangerous than direct misinformation attacks, and harder to detect.
- Threats get scored, not just described. CTSS brings quantification to a domain that has run on anecdotes.
- It’s the companion piece. Running AI SAFE² without CSF secures the tool and leaves the operator exposed. Run both.
Takeaways
- CSF v2.0: six resilience domains, 41-entry threat taxonomy with CTSS scoring, EFA/E7 protocol stack.
- Domain 6 (Digital & AI Symbiosis) is new. The human-AI relationship as a trained resilience domain.
- Edge cases documented honestly; dual-licensed for spread.
Bottom line: We spent decades learning to secure machines. v2.0 starts the work of securing the minds that use them. (Note: CSF v2.1 followed in July with the six domains fully operational, a follow-up post is queued.)
Release notes: Cognitive Sovereignty Framework v2.0 on GitHub, tag Cognitive_Sovereignty_Framework-v2.0.
