# Deepfake & AI Disinformation Ban


## Executive Summary
Generative AI now enables convincing audio, image, and video fabrications at scale. The harms include election interference, consumer fraud, reputational abuse, panic-inducing hoaxes, and non-consensual impersonation. This reform creates a narrowly tailored framework: banning malicious deepfakes in high-risk contexts, mandating clear disclosure for synthetic content in elections and public safety, and requiring platforms to maintain provenance standards and rapid takedown systems.

## The Problem

- **Elections at Risk:** Deepfakes threaten democratic processes by spreading false statements or fabricated videos of candidates just before elections.

- **Consumer & Public Safety Harms:** Fraudulent AI audio calls, scam solicitations, and fabricated emergency alerts can cause financial loss and public panic.

- **Reputational Abuse:** Individuals face impersonation or defamation through false depictions, without clear recourse.

- **Technology Gap:** While watermarking and provenance standards exist, they are not consistently used. Platforms lack uniform duties for labeling or incident response.


## The Reform

## 1. Core Prohibitions

- **Malicious deepfakes banned:** Illegal to create or distribute materially deceptive AI media in high-risk contexts with intent to cause harm.

- **Election-window rules:** Within 60–90 days before an election, distributing synthetic content about candidates or election processes is prohibited unless clearly labeled and not materially deceptive.

- **Robocall cloning ban:** AI-generated voices in robocalls are prohibited, aligning with the FCC’s ruling under the TCPA.

## 2. Labeling & Disclosure

- **Synthetic content disclosure:** Political ads, public safety alerts, and other high-risk uses must carry clear on-screen or audible disclosures plus metadata.

- **Provenance standards:** Major platforms must adopt C2PA-compatible content credentials and preserve provenance through upload and sharing.

## 3. Platform Duties

- Establish detection and response pipelines for high-risk deepfakes.

- Provide rapid takedown or corrective labels upon credible notice.

- Issue quarterly transparency reports.

- Maintain an appeals process for wrongfully flagged content.

## 4. Remedies & Penalties

- **Private right of action** for individuals and campaigns.

- **State AG and regulator enforcement** with civil fines.

- **Criminal liability** for willful, mass-scale disinformation causing election interference, fraud, or panic.

- **Expedited correction orders** requiring platforms to label or link to verified debunks.


## Safeguards

- **Protected Speech Preserved:** Parody, satire, and artistic expression remain lawful.

- **Narrow tailoring:** Restrictions focus only on materially deceptive content in high-risk contexts.

- **Due process:** Platforms must offer appeals, publish enforcement data, and face independent auditing.

- **No government editorial control:** Standards reference open technical protocols and independent audits, not government-approved content.


## Oversight & Auditing

- **Independent accredited auditors** verify compliance with provenance, disclosure, and takedown standards.

- **NIST & AI Safety Institute guidance** informs technical best practices, ensuring non-governmental but credible oversight.


## Technology Standards

- **Provenance:** Adoption of C2PA content credentials.

- **Watermarking:** Encouraged for AI generators where feasible.

- **Detection:** Continuous testing and process metrics, not mandated algorithms.


## Implementation Roadmap

**Phase 1 (0–6 months):** Definitions, disclosure templates, auditor accreditation.
**Phase 2 (6–12 months):** Platform duty-of-care systems for elections and public safety.
**Phase 3 (12–24 months):** Transparency reports, audits, harmonization with state election bodies and FCC enforcement.


## Integration with Other Reforms

- **Journalism Standards:** Verified outlets benefit from provenance tools and safe harbor when disclosing manipulations.

- **Social Media Transparency Rules:** Bot bans and ad registries complement disclosure of synthetic content.

- **Independent Oversight & Auditing:** Uses the same auditor ecosystem, reinforcing non-government accountability.


## Values Statement

This reform restores **trust, fairness, and accountability** by ensuring that AI-driven deception cannot undermine democracy, safety, or individual dignity.