Why AI Ghostwriting Feels Different — A Practitioner’s Deep Dive

Why this feels different

Ghostwriting has long been a quiet part of publishing and communications, and commentators have rightly pointed out the ethical grey areas that surround it (Inside Higher Ed). What feels different with AI is not just the output; it is a change in the sociotechnical system — the combination of people, practices, tools and institutions that make ghostwriting work. AI alters scale, speed, provenance and the locus of authorship. That matters to practitioners because it changes how you manage risk, craft agreements and deliver consistent quality.

What the evidence tells us

Professional bodies and reporting from writers’ gatherings have already started to translate these shifts into recommendations. The American Society of Journalists and Authors (ASJA) documented practical takeaways from a 2025 gathering: panelists encouraged starting small with AI — use it for tasks that add relatively little creative value but save time, such as transcription, brainstorming and summarising dense background reading. Separate ASJA guidance work has produced a statement of principles and a tiered approach to disclosure, recognising that different uses of AI merit different levels of transparency.

Surveys collected in that same ASJA work show mixed impacts: many writers find advantages in specific tasks while others report adverse economic and professional effects. Inside Higher Ed framed the comparison in moral terms: AI-assisted writing sits in the same ethical neighbourhood as traditional ghostwriting, but the social signals and expectations around authorship are shifting.

Three practical moves to adapt your workflow

• Map tasks to AI value: Create a simple matrix (low/high effort vs low/high impact). Use AI for low-effort, low-impact tasks first — transcription, initial research summarisation, title ideas and first-pass outlines (ASJA panel advice). Reserve high-impact creative choices — voice, narrative structure, argumentation — for human authorship or tightly supervised collaboration.
• Adopt tiered disclosure: Follow the ASJA approach of differentiating between purely utilitarian uses (transcription, grammar checks), substantive assistance (research summaries, draft expansions) and generative authorship. Make these tiers explicit in contracts and in conversations with clients so everyone shares expectations about provenance and responsibility.
• Build provenance and QA into the process: Capture prompts, model versions and sources used for any draft produced or heavily edited by AI. Treat AI output as a research artefact that requires verification. Establish human review checkpoints that focus on truth, voice, and ethical concerns (plagiarism, attribution). This preserves editorial control and makes accountability auditable.

Contracts, rates and conversations

Practitioners should translate technical choices into contractual language. If you permit AI use, specify which tiers are acceptable, who owns the prompt and output records, and who is responsible for fact-checking. Consider pricing differentials: tasks where AI materially reduces your time (for example, turning a transcribed interview into a first draft) could justify lower fees, but tasks that rely on human judgment and craft should retain market rates.

Maintaining craft and client trust

Use AI to augment not replace the aspects of ghostwriting that clients value most: voice, editorial judgement and the ability to hold nuance. That means making human responsibilities explicit in the workflow. For example, state that the human writer is responsible for final voice editing, factual accuracy and ethical choices — even when AI is used in drafting stages. That approach aligns with ASJA’s principles that frame writing as a fundamentally human activity and recommend disclosure to create mutual understanding.

Operational checklist for teams

• Decide permitted AI tiers and document them in one page of policy.
• Record model versions and prompts as part of deliverable metadata.
• Create mandatory QA checkpoints focused on fact-checking, legal risk and voice consistency.
• Train staff on the distinction between using AI for admin tasks and for generative content.
• Communicate with clients using plain language about where AI is used and why.

Final thought: design the sociotechnical system

AI changes the system around ghostwriting more than it changes the core job of telling someone else’s story. That shift invites deliberate design. Practitioners who treat AI as a component in a sociotechnical system — not as a magic box — can preserve craft, manage ethical risk and capture efficiencies where they matter. The current guidance from ASJA and reporting in trade and academic outlets give concrete starting points: start small, be transparent, and build QA and provenance into your workflow.

Sources

Comparing AI and ghostwriting — Inside Higher Ed

AI Is Changing Ghostwriting, and Other Top Takeaways from 2025 Gathering of the Ghosts — ASJA

ASJA Members Help Develop AI Guidelines for Ghostwriting — ASJA

AI is Affecting Independent Writers’ Work for Worse – and Better — ASJA

Expanded from

This post expands on a point from Sociotechnical systems in software and AI: Ghostwriting has always existed — why does AI ghostwriting feel different?.