The AI tool landscape changes so fast that recommendations from six months ago are often outdated. Here’s a current roundup of tools worth trying, based on my own testing and feedback from reporters actively using them.

I’m focusing on practical utility, not theoretical potential. These are tools that are helping journalists right now.

For Transcription: Whisper-Based Options

Whisper, OpenAI’s open-source transcription model, has become the foundation for numerous excellent tools.

MacWhisper (Mac only) runs Whisper locally on your computer. No upload required—the audio never leaves your machine. This matters for sensitive interviews. Accuracy is excellent for clear audio. The interface is simple. Cost: one-time purchase around $30 for the full version; limited free tier available.

TurboScribe is a web-based Whisper implementation with speaker diarization—it identifies who’s speaking. Useful for interviews with multiple sources. Accuracy remains strong. Free tier includes significant monthly minutes; paid plans for heavy users.

Otter.ai remains strong, though it’s not Whisper-based. The real-time transcription is useful for meetings or press conferences. The app handles recording and transcription together. Free tier includes 300 monthly minutes—enough for many freelancers.

My recommendation: if you interview sensitive sources, use MacWhisper or similar local tools. For general purposes, any of these work well.

For Research: The New Generation

Research assistants have improved dramatically. The current leaders:

Perplexity remains the best for cited research. Ask a question, get an answer with links to sources. The Pro tier allows deeper research with more complex queries. The free tier is genuinely useful for quick backgrounding.

I use it early in reporting—to identify sources, find background, discover angles I might have missed. Then I verify everything against primary sources. It’s not a replacement for research; it’s an accelerant.

Claude (Anthropic) has become my preference for working with documents. Upload a long PDF—a court filing, annual report, policy document—and Claude can summarize it, answer questions about it, and help you find specific information.

The free tier has limits but works for occasional use. If you regularly work with lengthy documents, the paid tier is worth it.

ChatGPT remains useful for explanation and ideation. When I need to understand a technical concept, ChatGPT can usually explain it at the right level. It’s also good for brainstorming angles, questions to ask sources, or ways to structure a story.

None of these should be trusted for facts. But they can help you find facts faster.

For Writing Assistance: Be Careful Here

I’m cautious about AI for actual writing—the quality concerns and detection issues are real. But some limited applications help:

Grammar and style tools like Grammarly’s AI features or built-in AI in Google Docs can catch errors and suggest improvements. I treat these like spell-check: useful for catching mistakes, not for generating content.

Headline brainstorming can work. If I’m stuck on a headline, asking Claude or ChatGPT for 10 options often produces a few worth considering. I never use their headlines verbatim—I use them as prompts for my own ideas.

SEO optimization tools can help with headlines and meta descriptions. Tools like Clearscope or Surfer SEO include AI features that help identify relevant terms. Use judiciously—SEO shouldn’t override editorial judgment.

What I don’t recommend: using AI to write article drafts, quotes, or any content that will be published under your name. The accuracy and quality issues are too significant.

For Data and Visualization

AI is increasingly useful for working with data:

ChatGPT Code Interpreter can analyze data files. Upload a spreadsheet, and ask questions about it—identify trends, create visualizations, calculate statistics. It’s not a replacement for proper data journalism skills, but it can handle quick analyses.

Flourish has added AI features for data visualization suggestions. Upload data, and it suggests appropriate visualization types. Useful for reporters without visualization expertise.

DataSquirrel is specifically designed for journalists doing data analysis. It can clean data, identify patterns, and generate summaries. Worth trying if you work with data regularly.

For Organization: AI-Enhanced Tools

Several organizational tools now include AI features worth trying:

Notion AI can help organize research, generate summaries of your notes, and create drafts of structured documents. If you already use Notion, the AI features are worth exploring.

Mem positions itself as an AI-native note-taking tool. It automatically finds connections between your notes and can answer questions about your collected materials. Useful for reporters juggling multiple ongoing stories.

Google NotebookLM lets you upload source materials and have conversations with them. It can also generate audio briefings—useful for reviewing material while commuting. Free for Google account holders.

What to Try This Month

If you’re just getting started with AI tools, I’d prioritize:

1. Pick a transcription tool. Any of the Whisper-based options will save real time.

2. Try Perplexity for research. Use it for backgrounding while remembering to verify everything.

3. Experiment with Claude or ChatGPT for document analysis. Next time you have a long document to work through, test whether AI can help.

Start with these limited applications. Get comfortable. Then expand based on what’s useful for your specific work.

Avoid trying everything at once—you’ll get overwhelmed and likely abandon everything. One tool at a time, integrated into your actual workflow.

The Trust Question

A note on trust: treat all AI output with appropriate skepticism.

AI tools hallucinate. They get things wrong confidently. They can’t replace human judgment about what’s true, what matters, and what’s ethical.

The tools I’ve described assist human journalists. They don’t replace verification, sourcing, or editorial judgment.

If you’re not prepared to verify everything AI tells you, don’t use it for journalism. The efficiency gains aren’t worth the accuracy risks.

Used carefully, these tools can make you more effective. Used carelessly, they can damage your credibility. The choice of how to use them is yours.