ChatGPT is OpenAI's assistant, and communications teams use it as the default drafting tool for volume work: press releases, pitch emails, social copy, first-pass research, and campaign scaffolding. Its value for PR teams is breadth: one subscription covers drafting, web search, image generation, and data analysis, with one of the widest install bases of any assistant, which lowers the cost of training a whole team on it. Two things changed in 2026 that this guide reflects: OpenAI retired several older models in February, and it is replacing custom GPTs with plugins.
This guide is the practitioner reference: the specific use cases, the prompts, the model and plan choices, the workflow patterns, and the risks for communications teams building ChatGPT into daily work.
Part of EPR's AI Communications Hub. Updated October 10, 2026. Model names, plan prices, and OpenAI policies change often, so confirm specifics on OpenAI's own pages before standardizing.
What does ChatGPT actually do for PR teams?
ChatGPT handles six recurring jobs inside a communications workflow: drafting press releases and pitches, brainstorming angles and headlines, summarizing research and data, generating and editing images, searching the live web for current information, and building first-pass campaign plans and content calendars. Communications teams typically pair it with Claude for careful long-form editing, Perplexity for sourced research, and Gemini for Google Workspace tasks. The four-tool comparison sits near the end of this guide.
Who should use this guide?
- PR agencies training a whole team on one shared tool for daily drafting.
- In-house communications teams producing high volumes of pitches, releases, and social copy.
- Marketing and content teams building campaign plans, calendars, and briefs at speed.
- Founders and executives drafting their own bylines, talking points, and social posts.
- Small teams and solo practitioners who need one tool that covers the most ground per dollar.
What is ChatGPT, and who built it?
ChatGPT is OpenAI's conversational interface to its GPT model family, launched in November 2022. OpenAI was founded in 2015 by a group that included Sam Altman and Elon Musk, and Altman has served as CEO since 2019. OpenAI has said that more than a billion people use ChatGPT every week, which makes it the most widely used assistant. For communications professionals, three characteristics matter most.
It is broad, not narrow. One subscription covers drafting, brainstorming, image generation, data analysis, and live web search. A team does not need five tools for five tasks, which is why ChatGPT is usually the first tool a comms team standardizes on, even when other tools get added later for specialized work.
It has a wide install base. Reporters, clients, and executives are more likely to already know ChatGPT than most competitors. That familiarity lowers training cost and makes it the safer default when rolling AI out to a team that has not used any assistant before.
It ships features fast. OpenAI adds agent features, app connections, and plugins on a rapid cadence, so the practitioner playbook changes faster here than on most other assistants. Check OpenAI's help center before building a workflow on any single feature.
Which ChatGPT model should a PR team use?
ChatGPT offers a fast default model, a slower reasoning model, and a top-tier Pro model, and the specific model names change every few months. OpenAI's help center says GPT-4o, GPT-4.1, GPT-4.1 mini, o4-mini and GPT-5 were retired from ChatGPT on February 13, 2026. Press reports say the GPT-5.5 family arrived in April 2026, with GPT-5.5 Instant becoming the default model for all users on May 5, and later coverage describes a GPT-5.6 family. Treat any model name as temporary and check the model picker before writing it into a policy.
| Model type | Best for in PR | Speed | Typical use |
|---|
| Default (Instant) model | Most day-to-day drafting, brainstorming and pitch writing | Fast | The default. Most comms work should route here automatically. |
| Reasoning (Thinking) model | Multi-step analysis, message stress-tests, and pressure-testing a position across many angles at once | Slower | Tasks where the team needs the model to reason through several steps, not just generate text. |
| Deep Research | Competitive audits and long research briefs with sources | Slowest | A report-style brief a professional then verifies. |
| Pro model | Long documents and the hardest analysis on Pro plans | Slow | Occasional heavy work, not daily drafting. |
Practical rule: let the default model handle most drafting. Reach for the reasoning model and Deep Research specifically for competitive audits, message stress-tests, and any task where the model needs to weigh multiple sources before answering, not just generate a paragraph.
Which ChatGPT plan should a PR team buy?
Most agencies should put client work on a Business or Enterprise workspace and keep personal Plus or Pro accounts for non-client work. Prices below come from third-party pricing trackers in October 2026, so confirm them on OpenAI's pricing page before budgeting. To judge whether a plan pays for itself, see How to Measure ROI on AI Tools for a PR Team.
| Plan | Approximate price | Fit for a comms team |
|---|
| Free and Go | $0 and about $8 a month in the U.S. | Light use with public information only |
| Plus | $20 a month | An individual professional working on non-confidential material |
| Pro | $100 or $200 a month | Heavy individual use and the top-tier model |
| Business (formerly Team) | $20 per user a month billed annually, $25 monthly, minimum two seats | Agencies and small teams that need a shared workspace and admin controls |
| Enterprise and Edu | Custom | Regulated clients and larger organizations |
Which ChatGPT surfaces matter for a comms team?
ChatGPT is a stack of surfaces, not one product, and each one changes what a team can do.
- chatgpt.com and the mobile apps are the surfaces most teams use daily.
- Web search gives live retrieval with inline citations, useful for fast fact-checks and current-events context.
- Agent mode completes multi-step browser tasks on higher paid plans, relevant for automating repetitive scheduling or research tasks.
- Plugins and connected apps are bundles of reusable instructions and app connections that replace custom GPTs, covered below.
- Business and Enterprise workspaces add admin controls, shared plugins, and data-handling terms that exclude training on conversations by default.
Are custom GPTs being replaced by plugins?
OpenAI says it is planning to retire custom GPTs and recommends moving workflows to plugins. According to its help center, new GPT creation and publishing are no longer available on personal Free, Go, Plus and Pro accounts, while Business, Enterprise and Edu workspaces can still create them where administrators allow it. For affected Enterprise workspaces, retirement is planned for December 11, 2026, and OpenAI expects other plans to follow the same timeline. Existing GPTs remain usable until retirement.
For agencies, that changes the old advice to build one custom GPT per client. The replacement is a plugin that holds the client's voice guide, banned language and output formats as reusable skills, with connected apps added only where needed. The setup steps, permissions and a migration timeline are in ChatGPT Plugins vs Custom GPTs for PR Agencies.
What are the best ChatGPT prompts for PR teams?
ChatGPT responds well to a clear brief: audience, goal, constraints, and source material. The prompts below are starting points, and the full library is in 25 ChatGPT Prompts for PR Teams, With Review Checks.
Press release first draft
"Draft a press release for a tier-1 business reporter. Brand: [name]. News: [the announcement in one sentence]. Key facts: [bullet list]. Quote from [name, title]: [quote or angle]. Boilerplate: [paste]. Constraints: 350 to 450 words, AP style, lead with the news, no marketing adjectives, no 'excited to announce.' After the release, list the three weakest points a skeptical reporter would push back on."
Pitch email to a specific reporter
"Write a pitch email to [reporter name] at [outlet]. Their recent coverage focuses on [paste headlines or topics]. The story I'm pitching: [one paragraph]. Why it fits their beat: [one line]. Constraints: under 150 words, no flattery, subject line under 60 characters, ends with one specific ask."
Brainstorm angles from a single fact
"Here is one fact about [brand or product]: [fact]. Generate 15 distinct pitch angles a reporter in [beat] would find newsworthy. For each, name the specific publication type it fits best and the one reason it would get passed over."
Content calendar from a campaign brief
"Build a six-week content calendar for [campaign name]. Goal: [one sentence]. Channels: [list]. Key dates: [list]. For each week, give the theme, three post ideas per channel, and the one metric that tells us the week worked."
Competitive research brief (use Deep Research or the reasoning model)
"Research how [three named competitors] have positioned themselves on [topic] in the past six months. For each, summarize their stated position, one specific example, and where it seems vulnerable to challenge. Flag anything you could not verify."
What does not work well on ChatGPT?
Three failure modes come up often enough to name directly.
Confident inaccuracy. ChatGPT can produce a fluent, specific-sounding statistic, quote, or source that does not exist. The fluency of the output carries no signal about its accuracy, so every fact, figure, and quote needs independent verification before it ships.
Voice drift on long documents. Without a strong voice sample loaded into a plugin skill or project, output on long pieces tends toward generic-professional phrasing instead of a specific executive's or brand's voice. For voice-locked work, many communicators prefer Claude, and the comparison table below explains why.
Treating memory as brand consistency. ChatGPT's memory feature is per-user, not shared across a team, and it does not train the underlying model. Asking it to "remember" brand guidelines across sessions is not a substitute for loading the guidelines into a shared plugin or workspace.
What are ChatGPT's data and confidentiality rules for PR teams?
Consumer plans can use conversation content to improve models unless a user opts out, while Business and Enterprise workspaces do not train on conversations by default and carry stronger data-handling terms. Brand, embargoed, or privileged material should go into Business or Enterprise workspaces only, never a free consumer account, and a team's data-handling policy should say so explicitly instead of leaving it to individual judgment. Confirm current terms against each client contract.
The legal stakes rose in 2026. In United States v. Heppner, a federal judge in New York reportedly ruled that a defendant's exchanges with a consumer AI chatbot were not protected by attorney-client privilege, so verify the current state of that case before relying on it. See AI and Client Confidentiality and AI Tools for Legal Communications Teams: Privilege Rules for the rules.
How does ChatGPT compare to Claude, Gemini, and Perplexity for PR work?
The four tools are not interchangeable. Each has a lane, and the right setup for most teams uses more than one.
| Tool | Best for in PR | Weakness | Typical PR use |
|---|
| ChatGPT | Broadest capability, volume drafting, brainstorming, the team default | Can confidently generate inaccurate specifics; voice drift on long pieces | First-pass press releases, pitch emails, content calendars, social copy |
| Claude | Long documents, careful tone, crisis statements | Tighter usage limits on standard tier | Editing 60-page reports, on-voice rewrites, message-house work |
| Gemini | Google Workspace integration, live search grounding | Less compelling as a standalone reasoning tool away from Workspace | Drafting inside Docs and Gmail, research in Sheets |
| Perplexity | Live, sourced research and competitive monitoring | Not built for drafting, synthesis only | Journalist background briefs, fact-checking, competitive brand monitoring |
A practical setup for most agencies: standardize the whole team on ChatGPT, add Claude seats for the people doing the heaviest drafting and editing, and add Perplexity for anyone doing research or monitoring day to day. For the deeper three-way breakdown, see Three Bots. Three Jobs. One PR Team.
How is ChatGPT changing what communications teams do?
The mechanical front half of the job compresses: the blank page, the slow first draft, the manual read-through of a long report for three relevant points. The parts of the job that still need a person do not go away: judgment on what the message should be, the relationships that earn coverage, and accountability for what ships under the brand's name. ChatGPT produces a draft, and a person owns it.
It also changes how brands are found. When ChatGPT recommends a competitor instead of your client, the cause is usually crawler access, entity clarity or source mapping, which Why ChatGPT Recommends Your Competitor: A 5-Test Diagnostic walks through.
Explore the cluster
