AI Training for Journalists & Newsrooms

A grounded look at what generative AI can and cannot do inside a reporting workflow, and where a hallucinated fact or invented quote can enter a story unnoticed. Participants leave with a shared vocabulary and a clear line between drafting support and reporting judgment.

• How large language models produce fluent text regardless of accuracy
• Where AI genuinely helps a reporting workflow, and where it does not
• Recognizing a plausible-sounding but unverified claim
• Decisions the journalist keeps
• Choosing a first, low-risk task to try on a live story

Reporters use AI to build an organized research plan from a beat or tip base, without letting a summary stand in for a document. The module builds a habit of tracing every research lead back to something checkable before it earns a place in a story.

• Turning a story idea into a research plan and question list
• Organizing public records and filings, with prior coverage, into a timeline
• Drafting background questions and follow-ups on named people or organizations
• Spotting gaps a summary glosses over and chasing them separately
• Keeping a research trail that shows where each fact came from

A practical workflow for recording and transcribing interviews. The organization step keeps the exact words a source used. Participants build interview guides and turn transcripts into usable notes while keeping the recording as the authoritative record.

• Building an interview guide from research gaps and a story angle
• Handling recording consent and notice, including on- or off-the-record terms
• Using transcription tools without letting cleanup drift from the exact quote
• Organizing transcripts by theme
• Flagging ambiguous or inaudible passages for a manual re-listen

Journalists use AI to organize the corroboration work around a claim, cross-referencing dates, names, and figures against primary documents, while keeping final verification a human judgment call. The module treats AI output as one lead to check against primary sources.

• Cross-referencing a claim against documents and records, alongside prior reporting
• Building a corroboration checklist for names, dates, figures, and quotes
• Using AI to flag inconsistencies without treating the flag as proof
• Distinguishing an unverifiable claim from a merely unfamiliar one
• Deciding when a claim needs a second source before it runs
• Invented citations, unsupported statistics

Reporters move from verified notes and transcripts to a structured draft that still reads like their own reporting while working through an editor's queries and revising a draft without reintroducing unverified material.

• Structure from verified notes
• Keeping a reporter's own voice and framing through AI-assisted drafting
• Responding to editor queries without losing sourcing detail
• Rewriting a passage for clarity without changing what a source said
• Deciding which AI-suggested phrasing to keep or cut, then rewrite

The closing module turns individual habits into a routine a newsroom or freelance reporter can run. It covers what never enters a consumer AI tool. It also covers corrections and what readers should be told about AI's role in a story.

• Keeping source identities and unpublished drafts, including documents, out of consumer AI tools
• Choosing approved tools and secure alternatives for sensitive reporting
• Running a correction once an AI-touched error reaches publication
• A plain-language disclosure practice
• Building a personal or newsroom checklist for source protection
• Planning a small pilot on one beat before wider adoption