A meeting anecdote, three lessons, an inspiring conclusion: leadership communication on LinkedIn already has its formulas. Artificial intelligence can reproduce them in seconds. It can also turn a unique experience into a text that any competitor could have signed. Looking ahead to September 2026, the communication challenge is becoming clearer: how can leaders publish more easily without becoming interchangeable? This forward-looking analysis draws on developments documented since the arrival of generative AI, without assuming which features or outcomes will actually be observed by that date.
The end of blank-page syndrome
For a leader, publishing involves more than writing. It means choosing a topic, finding an angle, checking information and anticipating reactions from employees, customers or investors. AI reduces some of this work: it suggests an outline, rephrases a voice note, shortens a technical explanation or offers several opening lines. A post that had repeatedly been put off can finally become publishable.
The trend extends beyond external tools. As early as 2023, LinkedIn announced AI-assisted writing features, notably for profiles and job listings, then tested assistance with drafting posts among Premium members. The platform also launched collaborative articles, initiated by AI and then enriched with human contributions. Assisted content production is therefore part of its evolution, not simply an unintended use by its members.
This ease of use can give a voice to experts who are uncomfortable with writing for publication, or to executives posting in a language other than their native tongue. AI-assisted writing is not necessarily less personal. It can even convey an idea better than a laboured draft. Everything depends on what is entrusted to the machine: the wording, or the thinking itself.
The risk: different voices, the same message
Generative models can produce plausible texts based on patterns they have learned. Given a vague instruction—write an inspiring message about leadership—they tend to deliver broadly agreeable ideas, tidy transitions and conclusions with no rough edges. This is not a malfunction: it is precisely what an instruction containing neither original experience nor a reasoned position calls for.
LinkedIn had its conventions long before ChatGPT: short sentences, stories of overcoming challenges, carefully managed vulnerability, a closing question designed to elicit comments. AI did not invent this grammar. It lowers the cost of reproducing it. The risk ahead is therefore a more crowded feed in which posts become difficult to tell apart, despite different photographs and bylines.
For a communications department, the trap is to confuse consistency with uniformity. One leader may discuss the energy transition, another recruitment and a third cybersecurity, while all adopt exactly the same narrative structure. Each text looks professional. Together, they erase the personalities, disagreements and distinctive ways of reasoning that make a genuinely personal voice interesting.
Expertise shows in useful details
In this environment, differentiation may come less from elegant wording than from the quality of the source material. Explaining that listening is essential adds little. Describing why a team abandoned a metric, how customer feedback changed a decision or which constraint delayed a project gives readers something to examine, discuss or even apply themselves.
Consider a fictional example: a leader wants to discuss the transformation of her customer service department. A generic post celebrates the complementary strengths of humans and machines. A genuinely informative post instead describes the decision to retain human approval for sensitive responses, the initial disagreement between two teams and the limitations that remain. It does not promise a universal formula; it explains the reasoning within a specific context.
This level of detail must remain compatible with confidentiality. It is not about disclosing customer data or featuring an employee without their consent. Anonymising information, obtaining the necessary permissions and clearly distinguishing a real case from a composite example remain essential. Authenticity does not require telling everything: it requires not fabricating what you tell.
Delegated writing, undiminished responsibility
Leaders did not wait for AI to work with ghostwriters, agencies or their in-house teams. A personal message is therefore not necessarily one written alone. It becomes credible when the person signing it recognises their ideas in the text, understands its nuances and can defend them elsewhere: in front of their teams, in an interview or beneath their own post.
AI undermines this contract if it invents an anecdote, hardens a conviction or turns hesitation into certainty. The leader then finds themselves endorsing an experience they never had, or a promise they never made. The problem goes beyond style. It concerns trust, especially when employees compare public statements with decisions made within the company.
An editorial workflow that protects the voice
- Gather material before drafting: start with an interview, a voice note or feedback from the field, rather than an abstract theme.
- Set limits on assistance: ask for a structure or rewording, without allowing the addition of facts, memories or quotations.
- Check, then seek approval: verify factual details and obtain the signatory’s agreement on the substance, not just the tone.
- Plan for the conversation: set aside time to respond to objections and explore points raised after publication.
Less focus on surface-level performance, more on trust
Measurement also deserves to evolve. Impressions and reactions indicate how widely a message circulates, but do not, on their own, establish its value. A post can attract considerable attention without strengthening its author’s credibility. Conversely, a low-profile piece of technical feedback can spark a useful discussion with a candidate, partner or customer. These effects call for qualitative assessment.
There is no basis for claiming that LinkedIn would systematically penalise a text simply because it was written with AI. Trying to anticipate an algorithmic penalty risks distracting from the real issue: usefulness to the reader. Likewise, a polished style is not enough to prove automated generation. Meaningful transparency is primarily about not misleading readers about the experiences described, the facts cited and the identity of the person taking responsibility.
What now? Looking ahead to September 2026, the most productive hypothesis is not that personal voices will disappear, but that their value will shift. As writing becomes more accessible, observation, judgement and responsibility could matter more. Communications teams would then be better served by collecting fewer post templates and documenting real decisions more thoroughly. AI can help find the words. It cannot have lived, on a leader’s behalf, the experiences worth recounting.
