I Used AI to Build a Newsroom With Fictional Reporters—and Kept a Human Accountable

Key points

• I directed an AI-assisted workflow to develop one current topic from each OMIKINA desk—AI, Robots, and Cybersecurity—into three original articles with source lists, illustrations, and podcast editions. Sources: S5, S6, S7, S8
• The desks are discovery systems. A feed headline or publisher summary can identify a question worth investigating, but it does not automatically become a verified OMIKINA claim. Sources: S1
• Clara Petra, Seth Stint, and Calder Rowe are disclosed fictional AI editorial personas. Their beats shape framing and questions; they are not human reporters, witnesses, credentialed experts, or independent sources. Sources: S2
• The linked records support factual statements; the selection, comparison, explanation, and conclusions are OMIKINA synthesis. Human direction, review, and public verification remain separate responsibilities. Sources: S1, S3, S4

I built a newsroom-shaped system, not a room full of people

I used AI to help build OMIKINA, monitor its desks, research current topics, compare sources, draft articles, create artwork, prepare podcast editions, implement the publication, and test the result. That is a substantial production system, but it is not an autonomous human newsroom. I define the assignment, approve the scope, correct the record, and remain accountable for what OMIKINA publishes.

The operating model is closer to a coordinated agent team. OpenAI describes a pattern in which a primary agent delegates parallel work to specialized subagents and synthesizes their results. In practice, I separate evidence research, skeptical review, writing, visual production, implementation, and verification so a persuasive draft is never the only test of completion.

Sources: S3, S4

Trending is a question, not an answer

I began with three live editions: AI, Robots, and Cybersecurity. They showed which subjects were moving across monitored publishers and official channels. Those signals shaped the questions, but they did not settle the facts. A trending headline can be timely while still being incomplete, promotional, mistaken, or unsupported.

The research therefore moved from discovery to evidence. Each synthesis had to trace important claims to primary or authoritative records, label company statements as company statements, preserve uncertainty, and avoid treating unlike measurements as comparable. The external sources provide the record. OMIKINA decides how to organize and explain that record, and that interpretive layer is labeled as synthesis.

Sources: S1, S5, S6, S7

The reporters are fictional; the accountability is not

Each fictional reporter received one topic that matched an assigned beat. Clara Petra took the AI policy and labor question. Seth Stint examined robotics models, grippers, and tactile hardware. Calder Rowe traced AI-enabled cyber risk through shared financial infrastructure. Their different lenses influenced emphasis; they did not change the evidence threshold or create firsthand reporting.

This first-person article carries the OMIKINA Editorial byline because the “I” is me, John N. Farmer—not one of the fictional reporters. AI contributed research, drafting, synthesis, imagery, audio preparation, implementation, and verification support. The personas do not conduct interviews or possess careers, credentials, relationships, private sources, or lived experience. I remain the accountable publisher.

Sources: S2, S5, S6, S7

Publication requires more than generated text

For each topic, the article, citations, image, audio file, narration text, metadata, feed entry, desk placement, and homepage treatment are separate deliverables. A completed draft does not prove that an image exists. A generated podcast does not prove that a browser can play it. A successful deployment does not prove that the correct story appears on every intended surface.

I count the work as published only after those public surfaces can be read back and the evidence boundaries remain intact. The three articles and their podcast entries are the current artifacts of that process. Readers can inspect their sources, see the fictional-persona disclosures, recognize the visuals as editorial artwork, and treat the written articles—not the AI narration—as the canonical record.

Sources: S1, S4, S5, S6, S7, S8

Why it matters

AI can help one publisher operate a broad, source-linked editorial system, but the trustworthy version depends on visible authorship, strict evidence boundaries, original synthesis, explicit fictional-persona disclosures, and verification of every published surface. The achievement is not making AI look human. It is making the production process inspectable enough that readers can see where the sources end, where synthesis begins, and who remains accountable.

Sources: S1, S2, S4

Sources

1. Editorial Standards — OMIKINA · 2026-09-01
2. About OMIKINA and AI Editorial Desk disclosures — OMIKINA · 2026-09-01
3. The builder’s guide to GPT‑5.6 — OpenAI · 2026-08-13
4. The Prompt Is the Org Chart: How to Give ChatGPT a Multidisciplinary Agent Team — OMIKINA · 2026-08-31
5. Congress Wants to Tax AI by the Token. The Hard Part Is Measuring Displacement. — OMIKINA · 2026-09-01
6. Smarter Brains, Uneven Hands: Why Physical AI Still Struggles at the Wrist — OMIKINA · 2026-09-01
7. The Patch Window Is Now a Financial-Stability Variable — OMIKINA · 2026-09-01
8. OMIKINA · Original Intelligence podcast feed — OMIKINA · 2026-09-01