The much-anticipated launch of GPT-5 massively increases the capabilities of AI. But what does this actually mean for PR, communications and corporate affairs practitioners?
I’ve been exploring it this morning and it lives up to most of OpenAI’s claims. It’s measurably faster, more accurate and far more context‑aware than previous versions. It has live‑data access and multimodal capabilities (text, images, audio) available in a single workflow. In practical terms, this means less time crafting prompts and more time using professional expertise and judgment.
OpenAI is the AI model powering Microsoft’s Copilot. It was fascinating to see Microsoft was far faster to deploy it than OpenAI. The live announcement said GPT-5 would be available immediately in most versions (including free) and next week for enterprise licences. In contrast in Copilot it started appearing immediately. I could see it in Microsoft 365 Copilot (its business and enterprise version) as OpenAI executives were still speaking at the launch.
No confusing choice of models
Copilot users also won’t see one of the big improvements, because they already had it. Using ChatGPT was confusing as it forced you to choose a model, all of which had baffling names like 4o, o3, o4-mini, o4-mini-high, 4.1, 4.1-mini. You’d have to be a real geek to know, or even care. Copilot users never had this confusing choice because it chose what it thought was the best model. For hard-core AI users this might be a disadvantage as you’re no longer immediately forcing it to use the ‘best’ model for different tasks. Copilot users will be used to the fact it often gives excellent answers, but occasionally rubbish ones because presumably it has chosen the wrong model. The solution is usually to try, try and try again.
Strategic context
GPT‑5 can ingest and reason over extraordinarily long inputs. Technically this is called extending its context window. Practically it means you can input and use far more information. Earlier models could ‘forget’ the first part of a long, complex conversation with lots of data. AI draws information from four sources – the user’s prompt, its large language model, searching the web and specific data and files the user shares with it. But there is a limit to how much data Or knowledge it can process all at one time.
If you like pointless jargon the last one about giving AI specific knowledge sources is technically known as RAG (retrieval augmented generation). It’s one of the most powerful ways that PR and comms can use AI. It helps ensure its answers and outputs are grounded in facts and reality.
This means you can upload campaign strategies, stakeholder maps, white papers, risk registers, government consultations, speeches, media coverage, annual reports and more to be used as a basis for the AI’s work.
Previously, you had to use clunky workarounds.
The result is far better written content with far fewer mistakes or hallucinations.
Multimodal workflows
Sorry, more jargon. Multimodal means it can handle text, audio, images and hopefully soon even video in one session. You don’t need to keep switching tools or modes. Practically, this could have multiple potential uses. For example, you could analyse text, image and audio content across multiple channels to check for messaging and brand consistency, accessibility and unconscious bias.
It should enable you to generate multimedia content from a single prompt. You could get it to brainstorm news angles and provide ideas for both headlines and images. These could be used as a basis for drafting a news release or briefing a client to show what an idea could look like.
Faster and smarter
Just exploring it this morning I’m struck by how much faster and smarter it is. First drafts appear faster and need far less tweaking. OpenAI claims GPT-5 is its “smartest, fastest, most useful” model and my experiments appear to confirm this. Deeper reasoning (making the AI think more about its answer), combined with live web access mean that drafts appear to be almost hallucination free. At last AI might stop making stuff up. Ironically, the downside to this might be that more mistakes creep through. If you’re fact checking and don’t spot many mistakes then human nature says you’ll eventually stop checking as thorougly.
From a PR and comms perspective it means it will be much better at tasks especially in fast-moving situations. For example in a crisis; using pre-approved positions, lines to takes, message houses and legal constraints; to quickly draft specific responses to situations for multiple audiences. It can be used to get to first drafts in a few minutes, rather than hours. The drafts can then be checked, edited and finalised by human experts.
Five examples of practical comms tasks for GPT-5
The power of GPT-5 opens up a realm of potential taks and activities that were theoretically possible, but difficult to do in practice.
Personal and executive personas
It’s already possible to build personal and executive personas to create AI agents that can write in the specific style and tone of an individual, either the user or an executive they needed to ghostwrite for.
Now that GPT-5 can handle far more information these personas can be trained on far more historic content and knowledge to make first drafts even closer to what the person would really say.
Crisis simulations
Because it can handle more information, faster it will be possible to run or test more advanced crisis simulations. Give it your crisis communications plan, relevant historical incidents, stakeholder maps, message house and previous statements to use as an accurate, relevant knowledge base. Then use it to run desktop simulations to game-plan new scenarios or stress-test planned scenarios.
Ask it to simulate counter‑moves from media, activists or political actors and stress‑test your planned responses, then produce board‑level options with risks and mitigations. The longer memory and improved reasoning make this radically more realistic than past simulations
Consistent messaging
Another benefit of handling more information is it should improve its ability to draft consistent messaging across complex projects over longer timeframes. Older AI models sometimes had to start from scratch, which meant a lack of consistency.
Data-driven insights
The new model is capable of analysing more complex data sets and using advanced reasoning to help interpret the data or do sentiment analysis.
Compliance checking
Copilot and ChatGPT were already capable of compliance checking to analyse content and check it against standards such regulations, guidelines, codes of conduct etc. It is far more capable as it can have a much larger knowledge base and is capable of advanced reasoning when doing the checking.
What’s next for AI and PR?
These are just some quick thoughts on how useful GPT-5 will be for PR and communications. The fact that it’s immediatelly available in Microsoft 365 Copilot is a huge benefit as it can be used in a safe, private and data secure environment.
I’d love to hear from you about ideas you’ve tried with GPT-5. What worked well and what didn’t? Let me know by getting in touch or commenting here.
We are prioritising updating our guidance and recommendations to existing clients. Over the next week or so we’ll be experimenting and researching more of the possible opportunities for PR and corporate affairs. Let me know if you’d like to discuss how these might apply to you.
