AI Can Write the Message, It Can’t Own the Consequences.
By Oluwayemisi Mafe
Artificial intelligence can draft a media release in seconds. It can turn a lengthy report into social media posts, suggest responses to difficult questions, summarize stakeholder feedback, develop key messages and produce the first draft of an executive speech before many of us have finished our morning coffee.
For communications professionals working under constant pressure to do more, faster and often with fewer resources, that capability is enormously attractive but there is a question we should be asking more often; when AI helps create a message and that message causes harm, who owns the consequences? The answer cannot be the technology.
An AI tool will not sit across the table from an executive team explaining why inaccurate information was published. It will not answer a journalist demanding to know how a false claim made it into a media release. It will not rebuild trust with employees after confidential information is mishandled. And it will not stand before stakeholders whose experiences were misrepresented by language that sounded perfectly polished but fundamentally misunderstood them, we will.
That is why the most important conversation about AI in public relations is no longer simply about what the technology can do. It is about what we remain accountable for when we use it. As much as the efficiency is real, so is the responsibility.
There is little value in pretending communications professionals should somehow remain untouched by AI. When used well, generative AI can be an extraordinary productivity tool. It can help us overcome the tyranny of the blank page, organize information, generate alternatives, identify gaps in a draft and adapt material for different channels and audiences.
The technology will continue to improve, and communications teams will continue to integrate it into their workflows. The question, therefore, is not whether AI belongs in communications. It is whether our governance, professional standards and judgment are evolving as quickly as our adoption of the technology. Truth is, there is a fundamental difference between delegating a task and delegating accountability.
We can delegate aspects of drafting to technology but cannot delegate our responsibility for what eventually carries an organization’s name. That distinction matters. A sentence can be grammatically flawless but factually wrong. A response can be perfectly structured and culturally tone-deaf. A stakeholder message can sound empathetic without demonstrating any real understanding of the people receiving it. An executive statement can be eloquent while making promises the organization cannot keep.
AI can produce plausible language remarkably well. Professional communicators must determine whether that language is true, appropriate, responsible and strategically sound. While accuracy is only the first test, since much of the discussion about responsible AI use focuses understandably on accuracy, generative AI can produce incorrect information with remarkable confidence, making verification essential. Factual accuracy is only one dimension of communications judgment.
Context needs to be considered as a technically accurate statement issued during a sensitive situation can still cause reputational damage if it ignores what stakeholders are feeling or experiencing. A message that works for one community may be inappropriate for another. Language that sounds neutral to an AI system may carry historical, cultural or political significance that changes how people interpret it.
Then there is confidentiality as well. Communicators routinely handle information that organizations would not want entered casually into third-party systems; unreleased financial information, employee matters, litigation, government relations strategies, crisis scenarios, patient or customer information, acquisition plans and executive deliberations. The convenience of asking an AI tool to ‘summarize this’ does not remove our obligation to understand where that information is going and whether we are authorized to put it there.
Bias also presents another challenge as AI systems learn from vast quantities of existing human-created material. They can reproduce assumptions, stereotypes and patterns of exclusion embedded in that material. A communications professional who accepts an output because it sounds polished may inadvertently amplify those problems.
And then there is something harder to measure; judgment. Should the organization respond at all? Is this the right moment? Who needs to be consulted before we speak? What might this sentence mean to someone outside the room? What are we missing? What happens if this message becomes tomorrow’s headline? Those are not mere questions, they are asking for strategy, relationships, ethics and organizational consequence.
The danger of confusing fluency with judgment is perhaps one of generative AI’s greatest strengths and also one of its greatest risks even when it sounds convincing. We have traditionally associated articulate writing with expertise. AI disrupts that assumption because it can produce authoritative-sounding language without possessing accountability, lived experience or organizational context. This makes human scrutiny more important, not less as a beautifully written answer can still be the wrong answer.
For communications professionals, this creates a new kind of risk. When an AI-generated draft is already polished, we may approach it as editors rather than interrogators. We correct a phrase, adjust the tone and perhaps verify a statistic but responsible use requires harder questions. Where did this claim come from? What assumptions underpin this recommendation? Whose perspective is absent? Is this consistent with what the organization has previously said and done? Could we defend every sentence publicly? And perhaps the most important question; would I be comfortable putting my professional judgment behind this? If the answer is no, the content is not ready simply because the technology says it is.
Human-in-the-loop cannot mean human-at-the-end. Many organizations respond to AI risk with the phrase ‘human in the loop.’ That is a useful principle, but only if the human is doing more than proofreading the final output. Human oversight should begin before the prompt is written. What information is appropriate to provide to the tool? What problem are we asking it to solve? Is AI even appropriate for this particular task? What expertise should inform the output? Who has authority to approve the final communication? These decisions are especially important when communications involve vulnerable communities, sensitive employee matters, public policy, health information, legal issues or crises.
In those situations, the communicator’s role is not simply to make AI-generated content sound better, it is to ensure the entire communications process reflects sound professional judgment that may sometimes mean deciding not to use AI at all. Professionalism has always included knowing which tools to use. Increasingly, it must also include knowing when not to use them.
Communications needs a seat at the AI governance table because responsible AI cannot be left exclusively to IT, legal or information-security teams. Those functions are critical, but communicators bring a different perspective. We understand audiences. We understand how language can be interpreted beyond its intended meaning. We think about trust, reputation, stakeholder expectations and the gap that can emerge between what an organization intends to say and what people actually hear.
That perspective belongs in organizational AI governance. Communications leaders should be helping organizations establish clear expectations around issues such as what information can be entered into AI systems, which communications require mandatory human review, how AI-generated facts and sources are verified, when disclosure may be appropriate, who approves high-risk content, and how errors involving AI-assisted communication will be handled. This is not about building bureaucracy around every prompt. It is about ensuring that the speed of adoption does not outpace the organization’s ability to manage the consequences.
Stop press, just before you press publish, perhaps responsible AI use in communications can begin with a simple accountability test. Before AI-assisted content leaves our hands, we should be able to answer these questions. Is it accurate? Have the facts, quotations, statistics and sources been independently verified? Is it appropriate? Does the message reflect the audience, cultural context, timing and circumstances in which it will be received? Is the information protected? Have confidentiality, privacy, copyright and organizational data policies been respected? Whose perspective might be missing? Have we considered bias, accessibility and the stakeholders most affected by the communication? Who owns this message? Is there a human being prepared to stand behind its accuracy, judgment and consequences? If nobody can confidently answer the fifth question, the first four are not enough.
Our value was never just the words; there is another reason I believe this moment should excite rather than frighten communications professionals. AI is forcing us to confront a question our profession has wrestled with for years, what is the strategic value of a communicator? If our value is primarily our ability to produce competent copy quickly, then increasingly capable AI presents an obvious challenge. But that has never been the highest value of our profession. Our value is understanding the environment in which words will land. It is recognizing the stakeholder who has not been considered. It is challenging an executive when the proposed message and organizational behaviour do not align. It is understanding that sometimes the right communications advice is not say this, but change this. It is anticipating how today’s statement may become tomorrow’s reputational issue. It is knowing when empathy is required, when evidence is required, when silence is appropriate and when an organization needs to listen before it speaks.
As AI helps us do some of our work faster, perhaps that gives us an opportunity to spend more of our time doing the work that requires precisely what AI does not possess; context, courage, empathy, ethical judgment and accountability. The future of communications, then, should not be framed as humans versus machines. The more useful distinction is between capability and responsibility.
While AI will become increasingly capable, when an organization communicates, someone must still decide whether the message is true, whether it is responsible, whether it reflects the organization’s values and whether it deserves the trust of the people receiving it. Truth is, AI can write the message but it cannot own the consequences.
Oluwayemisi Mafe, MCIPR, ACIM, MNIMN, ANIPR, is a strategic communications and public affairs leader with more than 15 years of experience helping organizations build trust, strengthen stakeholder relationships, and navigate complex reputation and policy issues across Africa and Canada. She specializes in corporate communications, public affairs, health advocacy, stakeholder engagement, and inclusive communications.