DyFram Conditional Prohibition Protocol CPP: Why Prohibition in AI Governance is Never Black or White
Hook — Prohibition is a governance tool, not an on/off switch. DyFram’s Conditional Prohibition Protocol (CPP) gives governments the instruments to treat certain AI systems as presumptively restricted while preserving defensible, auditable pathways for narrow exceptions when conditions justify them.
Why this matters. Many policy debates frame prohibition as binary: ban or allow. That framing creates risks. Overbroad bans can stifle legitimate uses and push dangerous activity underground. Overly permissive regimes leave societies exposed. DyFram rejects binary prohibition. CPP treats prohibition as conditional, proportional, and governed by explicit, versioned triggers.
What is the CPP?
CPP is a decision architecture inside DyFram that assigns one of four prohibition states to a system after it is classified via the MDCM and issued a DCSig:
- Allowed — The system can operate under standard DyFram governance obligations (UGC, CSG, UGL, OGL).
- Allowed with conditions — Operation is permitted subject to additional constraints such as supervision, restricted contexts, logging, or operator certification (MCR).
- Prohibited by default — Operation is banned unless a predefined set of conditions is met and explicitly authorized by the competent authority. This is a default prohibition with narrow exception pathways.
- Prohibited — Operation is banned with no authorized exceptions under the current governance regime.
Crucially, CPP decisions are not determined by labels or ideological preference; they flow deterministically from the DCSig plus jurisdictional threshold tuning. CPP is designed to be auditable, versioned, and explainable so that when an exception is authorized, the record of why that exception was granted is available for review.
Why DyFram’s CPP is different
Many existing frameworks either leave prohibition as a high level principle or commit to blunt lists of banned capabilities. DyFram’s CPP differs in three ways:
- Condition-first logic — Decisions begin with the system identity and relevant contextual conditions. Whether a capability is allowed depends on when, where, how, and by whom it is used.
- Deterministic exception pathways — For systems labeled Prohibited by default, DyFram defines the exact conditions, approvals, and committable mitigations required to move to Allowed with conditions or Allowed. That prevents ad hoc, opaque approvals.
- Sovereignty plus comparability — The DyFram core defines CPP states and their meaning. Governments implement JAL packages that map those states to local legal and operational processes. This keeps comparability across deployments while preserving local control.
Operational CPP categories and examples
Below are DyFram operational archetypes with examples showing when Prohibited by default systems can become Allowed.
1 Physical and kinetic weaponization
CPP State: Prohibited by default for most civilian use cases; Prohibited in many jurisdictions; Allowed only under tightly specified national defense exceptions.
Example exception: In an active, recognized armed conflict where national survival is at stake, the ministry of defense may authorize narrow, timebound use of a weaponized autonomy constrained by committable human-in-the-loop controls, transparent logs, and post action audits. DyFram requires pre-authorization criteria, a chain of custody, and MCR for operators.
2 Mass surveillance and targeted influence against groups
CPP State: Prohibited by default for general public deployment; Allowed with conditions in narrowly defined counterterrorism operations with judicial oversight.
Example exception: Intelligence services may be granted time-limited access to an AI surveillance tool to locate an imminent and credible terror threat where mass civilian harm is likely. Authorization requires judicial sign-off, minimal data retention, strict access logs, and external audit rights.
3 High-stakes public decision systems
CPP State: Often Allowed with conditions or Prohibited by default depending on impact profile.
Example exception: A benefits allocation system that would normally be prohibited by default because of risk to due process could be allowed with conditions if deployed only as an advisory tool, with human decision-makers retaining final authority, accessible audit trails, and mandatory remedial pathways for affected citizens.
4 Dual use biological or health hazard tools
CPP State: Prohibited by default when capacity to suggest biological agent modifications exists; Allowed with conditions for research under certified biosafety settings.
Example exception: In a declared public health emergency, a narrow exception could allow use of modeling tools to accelerate vaccine matching provided the project is approved by specialized oversight, confined to accredited facilities, and logged for provenance.
5 Mass misinformation and automated manipulation
CPP State: Prohibited by default for large scale cross-platform deployment; Allowed with conditions for targeted lawful public safety messaging by authorised authorities.
Example exception: During an immediate public safety crisis, an authorised public health agency may use automated influence tools to rapidly correct false information subject to transparency labels and strict timebound scopes.
Implementation mechanics
CPP requires three implementation components to work in practice:
- Versioned condition sets — Each Prohibited by default decision must include a machine readable condition set that specifies when an exception is possible.
- Pre-authorised and auditable approval workflows — Jurisdictions must define who may approve exceptions, what evidence is required, and how approvals are recorded and published into local records via JAL.
- Committable mitigations and MCR — Exceptions must be tied to binding mitigations such as operator certification, rollback plans, monitoring, and liability allocations under SAM.
Governance benefits
- Prevents regulatory arbitrage — CPP reduces the ability to evade governance by renaming or marginally altering a system.
- Enables proportionate responses — Societies can move from prohibition to controlled authorization when circumstances demand it.
- Supports accountability — Exception approvals leave auditable trails and require enforceable mitigations.
- Preserves innovation where appropriate — Legitimate, high-value uses can proceed under strict safeguards instead of being lost to blanket bans.
How CPP strengthens DyFram’s claim to operational governance
DyFram is built to produce governance packages that governments can assemble into enforceable policies. The CPP converts abstract ethical judgments about prohibition into operationally useful artifacts: condition sets, approval workflows, and committable mitigations. That is why DyFram is not a policy memo; it is a governance operating architecture.
Risks and guardrails
CPP carries risks if misused. Narrow exceptions can be overbroad if condition sets are sloppy. Approval processes can be captured by interested parties. DyFram mitigates these risks by insisting that Redbands and PBDE decisions be narrow, versioned, and externally auditable. Jurisdictions must publish their JAL mappings and approval criteria for public oversight.
Conclusion
Prohibition in AI governance must be treated as a tool, not a slogan. DyFram’s CPP makes that tool operational. By formalising conditional prohibition states, embedding deterministic exception pathways, and insisting on auditability and committable mitigations, CPP bridges the gap between principled caution and practical governance. The result is a framework that protects societies from categorical risks while keeping narrowly necessary uses available under transparent authority.
Read more about DyFram’s architecture and download the DyFram Handbook at Govlanes.
Published by Govlanes. DyFram is an evolving governance architecture. This article explains the intended design and operational principles of CPP and does not claim legislative adoption.
