AI for Journalism — Tools and Ethics for News, Reporting, and Media
Status: 🟩 COMPLETE 🟦 LIVING Section: decision-frameworks Tags: journalism, news, media, decision, ethics, reporting
The short answer
For Australian journalists, AI is changing the craft significantly:
- Strongly recommended uses: Transcription, research, document analysis, summarisation
- Use with care: Drafting, structuring, headline writing
- Avoid: Original reporting substitution, fabricating quotes or sources, AI-generated content without disclosure
Best tools: Claude or ChatGPT for general work; Perplexity for research with sources; Otter or Fireflies for transcription; Whisper for sensitive transcription (local).
Australian context: MEAA Code of Ethics applies; defamation law is strict; transparency expectations are increasing.
Where AI genuinely helps journalists
Transcription (the killer app)
Interview transcription was a major journalistic time sink. AI changes this:
Tools:
- Otter.ai or Fireflies.ai — for general use
- Descript — for podcast/audio production
- Whisper (open-source) — for sensitive material, run locally
Time saved: Hours per interview transcribed.
Workflow:
- Record interview (with consent)
- AI transcribes
- Search transcript for quotes and themes
- Quote verification still done manually
Document analysis
For investigative work with document dumps:
- AI can search and summarise large document collections
- Find specific information across thousands of pages
- Identify themes and patterns
- Help structure investigations
Tools:
- Claude (200K token context) for long documents
- Gemini (1M token context) for very long documents
- NotebookLM for research collections
- For sensitive: local AI (Ollama) on your own machine
Research and background
- Fast background on topics
- Multiple perspectives on issues
- Finding sources and related coverage
- Translation of foreign-language material
Best tools:
- Perplexity for cited research
- Gemini for current information
- DuckDuckGo AI for anonymous research
Summarisation
- Long articles to short summaries
- Multiple sources to overview
- Long videos/podcasts to text
- Government reports to readable summaries
Translation
- Translate foreign-language sources
- Multilingual interview subjects
- International coverage
Best tools:
- DeepL for European languages
- Google Translate for broad coverage
- Claude/ChatGPT for nuance
Data analysis
- Help interpret datasets
- Code generation for analysis
- Visualisation suggestions
- Statistical concepts explained
(For specific analysis: Python, R, or specialist tools with AI assistance)
Editorial preparation
- Brainstorming angles
- Outline drafts
- Headline variations
- Subhead suggestions
- Caption drafts
Routine writing
- Weather and traffic reports
- Sports score reports (with verified data)
- Earnings summaries
- Real estate listings articles
- Some classified content
What AI shouldn’t do
❌ Generate quotes from sources you haven’t interviewed
❌ Fabricate facts or sources
❌ Write articles without your reporting underneath
❌ Replace fact-checking
❌ Make ethical decisions for you
❌ Generate stories without disclosure
❌ Process newsroom-confidential material in free AI tools
❌ Be the source of “facts” without verification
The disclosure question
Australian newsrooms increasingly require disclosure of AI use:
When disclosure typically required
- AI-generated images in news
- AI translation of significant content
- AI summaries of complex documents
- Any audience-facing AI-generated text
- AI-augmented investigative work where AI played a substantive role
When disclosure typically not required
- Spell check and grammar AI
- Transcription tools (just a fact)
- Routine internal research
- Brainstorming and ideation
Disclosure formats
Common patterns:
- “This article was researched with AI assistance for [transcription/translation/summarisation]”
- “Image generated by AI” or “AI-generated illustration”
- “This story includes AI-translated quotes from [language]“
Major Australian news organisation policies
(Verify current — policies are evolving)
- ABC has detailed AI policies
- Nine Entertainment (SMH, AFR, Age) AI use guidance
- News Corp (The Australian, various titles)
- Guardian Australia
- The Conversation Australia
All have policies; specifics vary.
Australian journalism context
MEAA Code of Ethics
The Media Entertainment & Arts Alliance Code of Ethics applies:
- Honesty and fairness
- Acknowledge sources
- Don’t suppress relevant information
- Be transparent with readers
AI use must be consistent with these principles.
Defamation law
Australian defamation law is strict:
- AI-generated content about real people requires same verification
- Defamation suits can be expensive
- Don’t rely on AI for accuracy on individuals
Privacy Act
- Source confidentiality protected
- Personal information handling
- Sensitive information protections
Broadcast/online regulations
- ACMA standards for broadcasting
- Australian Press Council standards
- Various platform-specific considerations
For different types of journalism
Investigative journalism
Useful:
- Document analysis at scale
- Transcription of long interviews
- Pattern detection in datasets
- Translation of foreign-language documents
- Finding connections in complex material
Cautions:
- Source confidentiality requires AI tool choice with appropriate privacy
- Consider local AI for sensitive material
- Don’t rely on AI for verification
Daily news reporting
Useful:
- Background research
- Fast first-draft writing
- Summarising press releases
- Multiple-source synthesis
Cautions:
- Tight deadlines + AI = risk of error
- Always verify
- Don’t let AI become the source
Feature writing
Useful:
- Brainstorming approaches
- Structural suggestions
- Research breadth
Cautions:
- Voice is yours, not AI’s
- AI tends toward generic
- Reader expects your perspective
Photo journalism
Useful:
- Image organisation and tagging
- Editing assistance
- Caption help
Cautions:
- NO AI-generated images presented as photography
- AI image enhancement of real photos has limits (manipulation ethics)
- Photo selection remains journalistic judgment
Video journalism
Useful:
- Transcription
- B-roll suggestions
- Editing assistance (Descript-style)
- Caption generation
Cautions:
- AI-generated video footage is a serious ethical concern
- Disclosure required for any AI-generated visual elements
- Deepfake awareness
Broadcast journalism
Useful:
- Script preparation
- Background research
- Pronunciation guides
- Voice transcription for production
Cautions:
- AI voice clones of presenters raise questions
- Real voice for actual broadcast
Opinion and commentary
Useful:
- Brainstorming
- Stress-testing arguments
- Counter-arguments to consider
Cautions:
- Opinion is YOUR voice
- AI-written opinion is fundamentally inauthentic
- Readers come for your perspective
Sports journalism
Useful:
- Statistics and data
- Game summaries from box scores
- Background on athletes
Cautions:
- Routine reports OK with disclosure; analysis and feature work requires real reporting
Court and legal reporting
Useful:
- Document analysis
- Translation
- Background on legal concepts
Cautions:
- Strict accuracy requirements
- Contempt risks
- Defamation risks especially high
- AI hallucinations on legal specifics
The major ethical concerns
Fabricated quotes
Multiple high-profile cases of journalists submitting AI-generated quotes attributed to real people. Career-ending.
Rule: Never use a quote without recording or notes from the actual interview.
Fabricated sources
Similar: AI generates plausible-sounding sources. Don’t.
Rule: Real sources only; verify they said what you’re attributing.
Deepfake risks
- AI can generate fake videos of real people
- Awareness of detection methods important
- Verification before publishing any received content
AI-generated content presented as human
Discovery erodes trust in your work and the industry.
Rule: Disclose AI involvement.
Source confidentiality
If you process confidential source material through free AI tools, source confidentiality is at risk.
Rule: Use local AI or enterprise tools with appropriate agreements for sensitive material.
Bias propagation
AI training data has biases. AI-generated content may reflect these. Be aware.
Tools by use case
Best for transcription
- Otter.ai or Fireflies for routine
- Whisper local for sensitive
- Descript for production workflows
Best for research with sources
- Perplexity for cited results
- Gemini Advanced for current info
- Claude or ChatGPT for analysis
Best for long document analysis
- Claude Pro (200K tokens)
- Gemini Advanced (1M tokens)
- NotebookLM for research collections
Best for sensitive material
- Local AI (Ollama)
- Enterprise AI with appropriate DPA
- Never free consumer tools
Best for writing assistance
- Claude for thoughtful drafts
- ChatGPT for variety
- Specialist tools for specific tasks
Best for translation
- DeepL for European languages
- Google Translate for broader coverage
- Claude/ChatGPT for nuance
A reasonable journalist’s AI workflow
Daily reporter
- Claude or ChatGPT for general work
- Otter or Fireflies for interviews
- Perplexity for quick research
- Verify everything; disclose where required
Investigative reporter
- All the above PLUS
- Claude with long documents
- Local AI for sensitive material
- Strong verification discipline
Editor
- AI for editing assistance
- Summarisation of long submissions
- Pattern detection across submissions
- Quality control awareness
Freelancer
- Free tier of major AI tools
- Clear disclosure on submitted work
- Build AI awareness as a skill
Privacy in newsrooms
Australian newsrooms increasingly considering:
- AI tool selection with appropriate enterprise terms
- Source protection in AI workflow
- Subscriber data protection
- Editorial AI policies
For freelancers and individual journalists:
- Be aware which tools you’re using
- Don’t compromise source confidentiality
- Read tool privacy policies
The future of AI journalism
Trends:
- More AI in newsrooms for efficiency
- More AI literacy required for journalists
- Disclosure requirements increasing
- AI-augmented investigations becoming standard
- Audience expectation of transparency growing
- Specialist AI journalism roles emerging
For Australian journalists: AI literacy is no longer optional. The craft remains human; the tools are changing.
Common gotchas
- AI confidence ≠ accuracy. Always verify.
- Hallucinated citations are common. Every cite must be checked.
- AI-generated quotes are fabrication. Career-ending if discovered.
- Free AI for sensitive material risks sources. Use appropriate tools.
- Disclosure expectations vary. Know your masthead’s policy.
- Routine work risks reader trust. AI-written everything erodes connection.
- Defamation risks magnified. AI doesn’t know Australian defamation law.
See also
- ai-for-writers — broader writing context
- otter-ai — transcription
- fireflies-ai — alternative
- whisper — local transcription
- perplexity — cited research
- deep-research-mode — AI research
- hallucinations — verification context
- watermarking-ai-content — content provenance
Sources
- MEAA Code of Ethics
- Australian Press Council standards
- ACMA broadcasting standards
- Australian newsroom AI policies (ABC, Nine, News Corp, Guardian Australia, 2024-2026)
- Reuters Institute reports on AI in journalism
- ALM Practical Journalism resources
- Personal observation of Australian journalism evolution