The Most Common AI Mistakes PR Managers Make

ChatGPT and Claude will identify factually correct story angles because they can parse what your release contains. They cannot recognise which angle will make a specific editor at a specific publication decide to cover you instead of the three other releases in their inbox today. You end up with accurate but forgettable press releases that your media contacts ignore.

The fix

Write the core narrative angle yourself before you hand anything to AI. Use AI only to tighten language and check structure after you have already decided what story you are actually telling.

AI tools optimise for brevity and keyword density because those are measurable. A two-sentence AI summary of your company pivot removes the context that explains why this pivot matters to your industry or customers. Journalistss see the facts without the connective tissue that makes facts newsworthy.

The fix

Always read AI-compressed narratives through the lens of a journalist asking 'why should my readers care about this specific thing right now' and restore the context AI removed.

Claude and ChatGPT absorb your company voice from training data and your internal documents. They reflect back language that works inside your organisation but sounds like corporate messaging to journalists. The press release sounds polished to you and defensive to the people you need to read it.

The fix

Give AI a journalist article about a competitor or adjacent company as the style reference, not your own internal communications or previous releases.

Meltwater AI and Cision AI can identify journalists who cover your sector. They cannot tell you which journalists have editorial relationships with your competitors, which ones are working on a story that conflicts with your news, or which editor at a publication will actually push your story forward internally. You distribute broadly instead of strategically.

The fix

Use AI to identify the beat and publication, then apply your own relationships and recent coverage knowledge to choose three to five journalists you actually want to reach, not ten to fifteen that fit the algorithm.

Meltwater AI and Cision AI flag mentions and sort by reach and sentiment. They miss the tone shift where a usually supportive industry publication starts asking harder questions. They don't show you the journalist who has been quietly interviewing your competitors for a comparison piece. You believe you are not in a media conversation when you actually are, but in the wrong direction.

The fix

undefined

PR playbook templates built into Claude and ChatGPT say respond quickly, stay transparent, show empathy. These are baseline minimums. AI does not know if silence actually builds more trust in your specific situation, if your response will amplify a story that is currently contained, or if the people most affected by your crisis are not on social media. You end up following the playbook instead of reading the moment.

The fix

Write the first response yourself based on what you know about your stakeholders and the specific situation. Only use AI to pressure-test your judgement once you have made it.

AI tools will generate a single well-structured crisis statement that is technically accurate for employees, customers, regulators and the public. Each group needs a different message because each group experiences your crisis differently and has different information needs. Your one-size response satisfies no one.

The fix

Write your core position first, then create separate messages for each stakeholder group based on their actual relationship to the crisis, and only use AI to check consistency and tone across all versions.

Meltwater sentiment tracking and social listening show you volume and velocity of mentions. They do not show you whether waiting twelve hours for better information will change your response, whether responding now will set the terms of the conversation or whether the story will collapse on its own if you stay quiet. You release a statement too early and either have to walk it back or defend a position that facts undermine later.

The fix

Before you write any response, tell yourself the three things you need to know before you speak, check whether you know them, and commit to when you will respond regardless of media volume.

When you ask AI to analyse how competitor X handled a similar crisis, you get structure and templates. You do not get the specific conditions that made their response work or fail. Your situation has different stakeholders, different prior relationships, different regulatory context. Copying their language sounds like you are following a manual, not responding authentically.

The fix

Read what competitors did, note what worked or failed, then write your response as if you have never seen theirs, using your own stakeholder relationships and context to guide language.

Cision and Meltwater track major outlets, social platforms and blogs. They do not necessarily catch conversations in industry Slack channels, WhatsApp groups, employee forums or niche communities where your real stakeholders are actually talking about what happened. You believe the crisis is contained when the real conversation is happening where your monitoring tools cannot see it.

The fix

Ask your team where your customers and employees are actually talking about you right now, then check those specific places manually, and treat those conversations as your primary information source.

AI generates one email or pitch that works structurally for any journalist in your target list. It cannot know that one editor has been asking you for trends data for months and wants a different angle, or that another journalist is always more responsive to interviews than written materials. You send the same generic pitch to people who need different approaches.

The fix

Personalise the opening and the ask itself based on your actual history with each journalist. Use AI only for copy editing and structure, not for deciding what you are asking each person to do.

Claude will tell you a pitch is well-structured, grammatically correct and persuasive. It will not tell you whether the pitch arrives at the moment when a journalist is actively looking for stories like yours. You send polished pitches at the wrong time and get ignored, then blame the pitch instead of the timing.

The fix

Before you write anything, find out what beat your journalist is actively working on right now, pitch specifically into that moment, and only use AI to refine language once you know your timing is right.

When you use AI to monitor media conversations and draft responses, you stop having the phone conversations where journalists tell you what they are working on, what story they wish they could cover, or where your industry is actually heading. You gain reporting speed and lose the relationships that made your best placements possible. Your media mix becomes more democratic and less effective.

The fix

Use AI for media monitoring efficiency so you have time for phone calls instead of inbox triage. Call three journalists per week just to ask what they are working on, not to pitch anything.

Meltwater and Cision assign positive, neutral and negative sentiment to media mentions. These categories flatten the actual shape of a journalist's thinking. A piece marked 'neutral' might be far more damaging than a negative piece because it creates doubt instead of triggering a defensive response. You miss signals that matter because AI reduced them to a single number.

The fix

Read the actual articles that Meltwater flags as most important for your story, not the summaries, and decide for yourself whether the sentiment score matches the actual threat or opportunity in what was written.

When you ask AI which media mentions matter, you get algorithmic ranking based on outlet size and reach. An article in a niche trade publication read by your entire customer base gets ranked lower than a mention in a larger outlet your customers never see. You stop noticing the small conversations that actually change customer perception of you.

The fix

Tell your media monitoring tool to alert you on specific publications and beat reporters that reach your actual customers and decision makers, regardless of overall reach, not just the biggest outlets in your sector.