For PR and Communications Managers

40 Questions PR Managers Should Ask Before Trusting AI

AI can write a press release that reads correctly but fails to interest a single journalist. It can summarise media coverage accurately and still miss the story that matters. Your judgement about what will actually work in the real world cannot be automated.

These are suggestions. Use the ones that fit your situation.

Download printable PDF

Press Release Drafting and Distribution

1 Does this AI-drafted release have a single sentence that would make a journalist stop scrolling, or does it follow the standard structure that every organisation publishes?
2 Which specific journalists at target publications have covered similar stories, and does this release speak to what they actually wrote about last time?
3 Would I send this release to a relationship contact first, or would I be embarrassed to put it in their inbox?
4 Does the AI version explain why this matters right now, or does it explain what happened in the order it happened?
5 Are there quotes from the CEO or named executive that sound like they actually speak, or do they sound like a press release?
6 Has the AI included details that would interest a journalist but that the organisation didn't explicitly mention?
7 If I removed the quotes and company name, would a journalist recognise this as a real story or as corporate messaging?
8 Does this release answer the question a journalist would actually ask, or does it answer the questions the organisation wants asked?
9 Is there a specific trade publication that would care about this, and does the release use the language and context that publication understands?
10 Before distribution through Cision AI, have I tested this release on one real journalist contact to see if they would actually use it?

Media Monitoring and Sentiment Analysis

11 When Meltwater AI flags a mention as neutral, does the actual article contain criticism that a human reader would immediately notice?
12 Are there mentions the AI summarised that a competitor could weaponise, even though they read as positive in the summary?
13 Does the AI's categorisation of your brand sentiment match what you know about each journalist's actual stance on your organisation?
14 If the AI flagged a trend, can you trace it back to actual coverage, or is it pattern-matching across unrelated mentions?
15 Does the tone analysis from Claude or ChatGPT match what you remember about how that publication usually covers your sector?
16 Are there mentions the AI filtered out as low-priority that could signal an emerging story before it spreads?
17 When the AI summarises 50 articles into three key themes, which actual stories did it combine or drop?
18 Does the AI pick up on insider language or subtext that shows what journalists are really thinking, or just surface sentiment?
19 If your organisation is mentioned alongside a competitor, does the AI notice the comparison even if it's not explicit?
20 Has the AI identified which journalists are moving toward a critical stance before they publish a major story about your sector?

Crisis Communications and Response Strategy

21 When AI drafts a crisis response, does it sound like your CEO or organisation speaking, or like a legal template?
22 Does the AI response address what the public is actually angry about, or what the playbook says you should address?
23 Are there specific journalist contacts who broke related stories before, and does the AI version account for their particular interests?
24 If you publish this response, will it de-escalate the situation or become the story that spreads further?
25 Does the AI version acknowledge what people saw or heard, or does it jump straight to your explanation?
26 Are there trade-offs in this situation that the AI has missed because it optimised for sounding responsible?
27 Would you stake your reputation on this response, or does it read like something written to avoid liability?
28 Does the timing of publication matter in this crisis, and has the AI understood when the moment to respond actually is?
29 Are there stakeholders whose concerns this response ignores because they were not explicit in the input to the AI?
30 If this crisis touches on an existing relationship with a major stakeholder, does the AI version maintain that relationship or risk it?

Narrative Strategy and Relationship Judgment

31 When ChatGPT suggests a narrative angle, can you trace it back to what journalists in your sector are actually writing about?
32 Does the AI understand the difference between a story journalists will cover and a story your organisation wants told?
33 Are there relationships with specific journalists or editors that this AI-generated approach could damage?
34 Has the AI recognised which media contacts have editorial independence versus those who simply republish your statements?
35 When the AI suggests a particular angle, would the journalists you know actually find it interesting or just obvious?
36 Does the strategy account for how individual journalists have covered your organisation in the past, or treat all media as the same?
37 Are there storytelling choices in the AI version that could alienate specific communities or audiences you need to reach?
38 If you follow this narrative strategy, are you building the kind of media relationships that survive beyond the AI-assisted campaign?
39 Does the AI understand the context of your sector's current conversation, or is it working from generic communication principles?
40 What would a journalist who knows your organisation think if they realised an AI system had written this messaging?

How to use these questions

Related reads

The Book — Out Now

Cognitive Sovereignty: How To Think For Yourself When AI Thinks For You

Read the first chapter free.

No spam. Unsubscribe anytime.