The HUMAN AI Ethics Framework for Communications™
What you publish today shapes what AI presents as truth tomorrow.
A practical framework for embedding ethics, human judgment and accountability into every AI-supported communication workflow. It translates established guidance from PRSA, UNESCO, AAAI and NIST into five questions communicators can actually use.
The HUMAN AI Ethics Framework for Communications™
Ethics, Influence, Communication and the Construction of Trust
Langley Allbritton | AI Communications Consulting
The communicator's expanding influence
We are no longer communicating only with people.
Communicators have long understood how framing, repetition, emotion and authority shape belief. AI extends the reach of that influence well beyond the moment a message is published.
01
It becomes the record.
What we publish enters datasets, knowledge bases and institutional memory, shaping what AI systems later present as truth.
02
It travels without its context.
A message written for one audience may reach people who never saw the original, interpreted by systems that never knew why it was written.
03
It can trigger action.
Agentic systems now communicate, purchase, schedule and escalate based on what they ingest, often at machine speed and with little human review.
Because we understand the mechanics of influence, communicators must serve as ethical guardians of the information environment.
The framework at a glance
Five questions, applied before, during and after AI-supported work.
HUMAN makes ethics a repeatable part of workflow design rather than a last-minute check.
H
Human purpose
Should AI be used? How does this impact trust?
Define the legitimate purpose. Ask whether AI is necessary, proportionate and credible for the task.
U
Understand people and inputs
What enters the system? Who will it impact?
Identify stakeholders and minimize sensitive data. Confirm consent, provenance and ownership before prompting.
M
Maintain human judgment
What must a person decide?
Name the accountable owner. Reviewers must be able to challenge, change, reject and escalate.
A
Assess truth and impact
What could go wrong?
Verify facts, sources and quotes. Test for bias, exclusion, manipulation and foreseeable misuse.
N
Name, navigate and notice
What must be disclosed, recorded or reported?
Disclose material AI use and document key decisions. Monitor after release and correct quickly.
Where each checkpoint enters the workflow
Before use
H + U
Justify the use, classify the risk, choose an approved tool and set data boundaries.
While creating
U + M
Constrain the prompt, protect information and keep a real human decision-maker.
Before release
M + A + N
Verify, test impact, approve, disclose when material and keep a record.
After release
A + N
Monitor, receive challenges, correct, and update the workflow with what you learned.
Guardrails
Match the level of oversight to the stakes.
Not every task needs the same review. Classify the work by its consequences, sensitivity and reversibility, then raise safeguards as risk rises.
GREEN · Routine
Brainstorming, outlining, rewriting non-sensitive copy, summarizing approved internal material.
Safeguard: approved tool, no sensitive data, practitioner owns the result.
YELLOW · Material
External claims, executive voice, media materials, crisis drafts, content about identifiable people.
Safeguard: source verification, second-person review, documented owner and disclosure decision.
RED · Prohibited
Synthetic impersonation, fabricated quotes or support, undisclosed bots, autonomous crisis response.
Safeguard: do not proceed without formal legal, ethics, security and executive authorization.
“A communicator must not cause a machine to do something the communicator could not ethically or lawfully do personally.”
The complete guide
The tools to put it into practice.
This page shows the model. The full 30-page guide gives your team the assessments, checklists and instructions to run it on real work.
Request the complete guide (free)AI Code of Ethics
Ten commitments drawn from PRSA, UNESCO and AAAI.
AI Communication Impact and Influence™ Chart
Seven stages from publication to long memory.
Blurred lines
What “it’s only a draft” and similar phrases can hide.
Ethics instruction set
A starting prompt for custom GPTs and AI agents.
Agentic and physical AI
Considerations when information turns into action.
Ethical measurement
Measuring belief, decisions and machine representation.
Communications Impact Assessment
Eight dimensions ending in proceed, modify, escalate or stop.
The Future-person test
Will someone years from now understand its context and owner?
Put it to work
Bring the framework into your organization.
Communications teams
Build HUMAN checkpoints into your approval flow so ethics happens at the first prompt, not the final review.
Agencies
Give clients a clear, defensible answer when they ask how you use AI on their behalf.
Associations and events
Workshops and keynotes on ethics, influence and trust for members navigating AI adoption.
Built on established work
HUMAN is an original, communications-specific synthesis of established ethics guidance. Every underlying principle is credited in the guide’s source map.
Currently under review by PR industry associations as they develop AI guidance.
Developed by Langley Allbritton, founder of AI Communications Consulting, August 2026.
The question behind every message