Every few months, someone posts a LinkedIn thread about AI ghostwriting being fundamentally dishonest. The argument goes: if you did not write it, you should not put your name on it. The thread gets thousands of likes, mostly from people who have never had to maintain a consistent publishing cadence while running a company.
This article takes the ethics question seriously. The counterargument has real weight, and founders who use AI assistance should be able to articulate why they think it is acceptable, not just dismiss the concern as performative. But the ethics conversation is often framed badly, and a poorly framed question produces a poorly considered answer.
Ghostwriting Has a Long Track Record
Most memoirs attributed to politicians, athletes, and CEOs were written with substantial help. The autobiography market has operated on ghostwriter collaboration for decades. Business books that shaped how entire industries think were produced through intensive collaboration between the named author and a writing partner who handled structure, prose, and often the research. The named author supplied the thinking, the experiences, and the positions. The writer supplied the craft.
This arrangement is widely accepted because the principle it rests on is clear: if the ideas and lived experience belong to the person whose name is on the cover, the writing assistance is legitimate. Nobody reads a CEO's memoir and expects it to be entirely self-produced. The implicit contract is about the substance, not the production method.
The discomfort around AI assistance on LinkedIn is worth examining separately, because the platform does carry a somewhat different expectation. But the starting point should be honest: the idea that writing without assistance is somehow more virtuous has a very thin historical foundation.
What the Counterargument Actually Says
The honest version of the ethics concern is not "you should write every word yourself." It is something more specific: LinkedIn carries an implicit promise of direct voice. When someone reads your post, they are assuming it represents how you actually think, what you genuinely believe, and roughly how you would put it if they asked you directly in a conversation. Content that fundamentally misrepresents your views, or that uses a register so detached from your natural style that it could have come from anyone, would violate that expectation.
This is a real concern. It deserves a real answer rather than a dismissal. If your posts start sounding like a different person entirely, or if they express positions you never formed and could not defend in a conversation, that is a trust problem. Not a legal problem. Not even necessarily a detectable problem. But a trust problem that erodes the actual value of having a presence on the platform.
The counterargument is strongest in one specific case: when AI generates ideas the founder then claims as original. If you are prompting with something like "write me a thought leadership post about supply chain resilience" without any underlying position of your own, you are outsourcing cognition, not just production. That is the thing genuinely worth worrying about.
Where the Actual Ethical Line Falls
The line is not at AI assistance. The line is at originality of thought.
A two-minute voice note is not a neutral prompt. It is an argument, an observation, a story fragment, or a position you arrived at through your own experience. You had that thought before you opened any app. The structural and editorial work, converting that raw material into a well-formed post, is the part you are getting help with. The AI does not invent your reasoning. It does not insert claims you did not make or positions you did not hold.
We are not saying this distinction is always clean or that every use of AI assistance is identical in ethical weight. A founder who records a genuine two-minute opinion and reviews the draft before publishing is doing something meaningfully different from someone who gives a topic keyword and publishes whatever comes back without reading it. Both might be using the same tool. The process matters for whether the content actually represents you.
You can read more about how Curioeus handles the voice-to-draft workflow if you want to see what that process looks like in practice.
The Transparency Question
Should you disclose that you use AI assistance for LinkedIn posts? The current answer is that there is no norm on the platform requiring this, and the people most vocal about demanding disclosure are often themselves publishing content that passed through PR teams, communications coaches, or executive assistants who substantially shaped the writing.
The more practical test: would you be comfortable if a reader knew the post started as a voice note that you recorded, reviewed, and approved before publishing? If yes, the process is defensible. If something about the process makes you uncomfortable when you describe it out loud, the issue is probably the process itself rather than transparency about it.
There is a version of AI assistance that would warrant disclosure or concern: using it to write about experiences you did not have, positions you do not hold, or expertise you are fabricating. That is a different activity, and the ethics problem there is not the AI, it is the fabrication.
Calibration Drift: The Practical Risk
The most concrete risk in AI-assisted writing is not one anyone talks about much: calibration drift. Over time, if you approve drafts without genuine engagement, the model's understanding of your voice starts being shaped by what you accepted, not by how you actually think and speak. Posts begin to drift toward a slightly more polished, slightly more generic register than your real voice carries.
The edit step is not a formality. It is where you catch the sentences that could belong to anyone, where you push back on structure that feels too tidy to be authentic, and where you add the specific detail that only you would know. This is also how the calibration improves over time rather than drifting. Founders who treat the approval loop as a rubber stamp tend to end up with content that technically came from their voice note but no longer sounds like them after three months.
The Founders Who Should Pause
There is one category of founder for whom the ethics question genuinely requires more thought: founders whose personal brand is explicitly built on radical authenticity. If your positioning is that you publish raw, unfiltered thinking that no other exec in your space is willing to say, your audience is holding you to a higher standard of directness than most. That does not make AI assistance impossible. It means your edit threshold needs to be much higher and your voice notes need to be genuinely unfiltered starting material.
For most B2B founders, the audience is not performing an authenticity audit. They want to know whether you understand the problems they face, whether you have relevant experience, and whether you say things worth returning for. Those qualities come from the ideas and the positions, not from whether you drafted in a notes app or spoke into your phone.
A Test Worth Running
Before publishing any AI-assisted post, try reading it out loud. If it sounds like you talking, the calibration is probably right. If it sounds like a very polished version of someone who thinks similarly to you, it is one edit round away from being right. If it sounds like nobody in particular, send it back and push for a rewrite that starts from more specific material from your original recording.
This is less a moral test than a quality test. But it tends to answer the ethics question at the same time. Content that reads as genuinely yours is also content you can stand behind. Content that reads as generic is both lower quality and harder to defend as representing your actual views.
The ethics conversation is worth having seriously, not dismissing. The answer is not that AI assistance is always fine. The answer is that the ideas and the positions have to be real, the approval process has to be engaged, and the result has to be something you would say. See the discussion of authentic voice in AI-assisted writing for more on how this plays out in practice, or check our about page for how we think about the product we are building.