Artificial Intelligence, Synthetic Media & Automated Accounts
ARTIFICIAL INTELLIGENCE, SYNTHETIC MEDIA & AUTOMATED ACCOUNTS
AI, TECHNOLOGY & CYBERSECURITY
America should lead the technological future.
MPA supports:
Critical-infrastructure cybersecurity
Identity protection
Deepfake protections
AI impersonation protections
Intellectual-property protection
Children's online safety
American semiconductor capacity
Quantum research
Advanced computing
When government uses AI to make consequential decisions affecting liberty, benefits, taxation, employment, policing, or similar matters, a human government official must remain accountable.
Government should never be able to say:
"The algorithm decided."
Artificial-intelligence policy should also address safety and accountability, employment displacement, personal-data use, training-data and copyright rights, deepfakes, election manipulation, online surveillance, cybersecurity, critical infrastructure, and responsible government use of automated systems.
AI, Fake Content, and Information Integrity
Automated Accounts
A public-facing account substantially controlled by software must disclose that it is automated when it distributes messages, comments, endorsements, or other communications at scale. Platforms must provide a visible automated-account label and a reporting process for undisclosed coordinated bot networks. The rule does not require people to disclose ordinary spell-checking, research assistance, accessibility tools, scheduling tools, or limited AI assistance used by a real person.
Synthetic Media
Materially AI-generated or altered audio, video, or images must carry a clear disclosure when a reasonable person could mistake the content for an authentic recording and it is used in a paid advertisement, commercial endorsement, impersonation, purported evidence, or election communication. Satire, parody, art, accessibility uses, and clearly fictional entertainment remain protected when they do not falsely claim to be authentic evidence or another person's real statement.
Fraud, Impersonation, and Exploitation
AI use does not create a new censorship power over opinions or disputed claims. Existing and updated law should impose civil and criminal liability when synthetic media is knowingly used for fraud, extortion, identity theft, fabricated evidence, nonconsensual sexual exploitation, or a false commercial or political endorsement. Affected people must have an expedited process for removal, correction, preservation of evidence, and recovery of provable damages.
Government Record Authentication
Federal agencies must cryptographically sign official digital documents, photographs, audio, and video when authenticity is material and provide a free public verification service. The absence of a signature does not by itself prove that content is false. Agencies must retain the original record, document later alterations, and preserve public-record and evidentiary requirements.
Enforcement
The Federal Trade Commission and state attorneys general may enforce commercial disclosure and deception rules. Civil penalties should increase for repeated undisclosed automation, coordinated impersonation, large-scale consumer harm, or deliberate removal of required provenance information. Political-speech rules must be viewpoint neutral and may regulate deception and disclosure, not lawful opinion.
WHY THIS MATTERS
AI can expand productivity, science, medicine, education, and national defense, but it can also concentrate power, displace workers, manipulate evidence, invade privacy, and impose large energy and water costs. America should lead without giving government or corporations unreviewable power over people.
WHAT MPA WILL DO
Require notice, meaningful explanation, human review, and an appeal whenever government uses automated systems for benefits, taxes, employment, licensing, policing, sentencing recommendations, or other consequential decisions.
Establish clear liability for fraudulent AI impersonation, malicious deepfakes, automated discrimination, reckless deployment in safety-critical systems, and concealment of known material risks.
Require provenance or durable disclosure for synthetic political advertising and official government communications, without creating a general licensing system for speech.
Protect personal data and intellectual property through consent, disclosure, security, audit, and lawful compensation rules appropriate to the use.
Condition major data-center incentives on public cost-benefit analysis, enforceable job and infrastructure commitments, resource-use reporting, and clawbacks.
Invest in grid capacity, semiconductors, cybersecurity, advanced computing research, and worker retraining tied to demonstrated labor-market effects.
WHAT MPA WILL MEASURE
Automated-decision error and appeal rates; confirmed discriminatory outcomes; security incidents; worker displacement and retraining results; domestic semiconductor capacity; data-center electricity and water intensity; residential rate impacts; promised versus delivered jobs; and incentive dollars recovered through clawbacks.
HIGH-RISK AI, SURVEILLANCE & AUTONOMOUS WEAPONS
Government use of AI does not create an exception to the Constitution. Persistent tracking, facial recognition, mass profiling, or analysis of private information must satisfy the same legal standards that would apply if a government employee performed the action directly.
High-risk automated decisions involving liberty, benefits, taxes, employment, healthcare, credit, education, policing, or government services must provide notice, an understandable explanation, an accountable human decision-maker, and a meaningful appeal.
A person challenging a high-risk automated decision must receive a written human decision within 30 calendar days, an independent appeal decision within the next 30 calendar days, and any complaint or independent review within the next 30 calendar days. On day 91, administrative remedies are exhausted and judicial review is available. Immediate court access remains available when delay would cause irreparable harm.
AI controlling critical infrastructure must include:
Independent security testing.
Audit logs.
Human override.
Safe failure modes.
Manual operating capability.
A physical disconnection or shutdown method where technically appropriate.
No AI system may independently authorize a nuclear launch. Rules for other autonomous weapons must preserve accountable human command over intentional lethal force, while permitting properly limited automated defense against incoming weapons when reaction time makes direct control impossible.
CHILDREN'S ONLINE PRIVACY & PROTECTION
Online services reasonably likely to be used by minors should provide the strongest privacy settings by default.
They may not knowingly:
Sell a minor's personal information.
Use precise location for advertising.
Target minors using sensitive personal profiles.
Employ manipulative designs intended to defeat a child's expressed choice.
Retain a minor's data longer than reasonably necessary.
Facilitate grooming, sextortion, trafficking, or sexual exploitation.
Parents and guardians should receive practical controls, but protecting children must not require every American to upload government identification or submit to facial recognition.
Age-assurance systems must minimize data, avoid permanent identity files, and provide alternatives for people who cannot safely or reasonably use one method.
Related policies: Data Centers & Digital Infrastructure; National Cybersecurity Standards & Trusted Personnel; Personal Data & Records Rights; Workers' Rights; Image, Likeness & Digital Replicas.
Status: Proposed / Draft — not yet formally adopted.
Originally published: August 24, 2026.
Last updated: October 2, 2026.
Version: Draft 0.8.