It’s fast-tracking research, planning and analysis. It’s kickstarting presentations, campaign reporting and content ideas. AI is getting a constant avalanche of more information than it needs to complete a task. A whole spreadsheet when only a few fields matter. A full client brief when only one section is relevant.
The people who get the most from AI aren’t necessarily the ones who give it the most. They know how much context to provide, what to leave out and how to add a human touch.
Here are seven things worth checking if you’re using AI at work.
1. Check your setup
Not all AI accounts handle information in the same way. Personal and business plans can have different data settings, controls and terms.
Review:
- Your account: When using ChatGPT at work, is it via your personal account or business workspace?
- Approved tools: If your workplace has adopted an AI platform, stick to it for consistency, security and control.
- Confidential information: Before sharing sensitive material with ChatGPT or other platforms, ensure the terms and any relevant agreements allow it. If you're unsure, get approval first.
- Data settings: Check whether your chats can be used to train AI models. ChatGPT Business and Enterprise exclude workspace data from training by default, while other accounts and tools may require you to adjust your settings.
2. Stop oversharing
It’s tempting to export the whole spreadsheet and let AI sort it out. It’s also becoming increasingly common. A 2026 PagerDuty survey found that 40% of Australian office professionals surveyed had entered customer data into public AI tools. Most tasks can, and should, be done without sharing confidential material.
Let’s say you want AI to analyse last quarter’s campaign performance. It needs spend, channel, leads and conversions. It doesn’t need customer names, phone numbers, email addresses or 40 unrelated CRM fields that came along with the export. Strip out anything the task doesn’t require before you share it. You’ll reduce unnecessary exposure and give AI a cleaner set of information to work with.
3. Keep the keys out
Some information should never go into a prompt, whatever tool or account you’re using:
- passwords
- API keys and authentication tokens
- recovery codes
- private keys
- full banking or card details
It happens more than people may think. An error message includes a login. A troubleshooting prompt includes an API key. A setup document may still include credentials.
If you need to share the surrounding information, remove the sensitive details first. For example, rather than pasting the actual API key, replace it with [REDACTED].
4. Check your connections
You may have granted access to more business information than you realise. Connecting to email, Google Drive or Dropbox, calendars, Slack or Microsoft Teams and CRM platforms can make daily work easier, but access may remain active long after it's needed.
Every quarter or so, review your connected apps and disconnect any you no longer use.
5. Manage the memory
Many AI tools can remember information between conversations, including your role, business, preferences and projects you’ve discussed. That can make them more useful because you don’t have to start from scratch each time.
It’s still worth checking what history it retains. Review your memory and personalisation settings from time to time, and keep them up to date as projects evolve, internal conversations happen and client feedback is received.
6. Use as a guide, not gospel
ChatGPT comes with a literal disclaimer: “ChatGPT can make mistakes. Check important info.” The National AI Centre found that only about half of Australian SMEs using AI have a process for checking its output before it affects customers. The judgement behind what gets used, changed and shared still belongs to people.
AI can help teams:
- research faster
- analyse more information
- generate ideas
- develop first drafts
- explore different approaches
- cut down repetitive work
People still need to take responsibility for the result:
- Accuracy: Proofread and fact-check the output. AI can sound confident and still be wrong.
- Context: Check the output against current briefs, supporting materials, email threads and what has been agreed.
- Strategic judgement: AI can process information and suggest directions, but a person still needs to provide commercial understanding, priorities and nuance.
- Brand and voice: Great writing isn't just grammatically correct. It needs to sound like the brand, talk like a human and be wordsmithed to improve flow, rhythm and impact.
- Creativity: AI can generate plenty of ideas. People still need to challenge them, solve problems creatively and avoid the snooze-worthy sameness that comes from accepting the first plausible answer.
- Relevance: AI has a particular talent for unnecessary waffle. It takes a human to focus on what matters to the audience, refine the message and communicate it clearly and concisely.
- Final decisions: Ensure someone with the relevant expertise has reviewed the work and is accountable for what reaches the client.
7. Get your team on the same page
Only half of Australian employees report having clear guidance, training or support for using AI, according to Gartner. A collaborative approach makes it easier for everyone to use AI to its full advantage. It gives your team a consistent starting point, so everyone knows what’s expected and works to the same standard.
A business AI policy doesn’t need to be complicated. A one-page internal guide could cover:
- Approved tools: Which platforms can be used for work.
- Purpose: Where AI can help and how it should be used.
- Information: What should never be shared.
- Confidentiality: How client and business information should be handled.
- Human review: What needs to be checked, refined or approved by a person.
- Accountability: Who is responsible for the final result.
- Support: Who to ask when someone isn’t sure.
How PIER uses AI
These tools are now part of how we work at PIER, but the bigger conversation is how they can add even more value by removing friction, streamlining processes and connecting ideas, information and expertise.
Our team uses AI for research, analysis, drafting and idea development to move faster and explore more options. We’re also paying close attention to how AI is changing and people search for and discover information, as we discuss in our guide to AIO, AI SEO and AEO.
The trick is to use AI thoughtfully, responsibly and to your advantage. Keep the strategy, creativity and accountability for the people who bring the work to life.
Frequently asked questions
How should businesses use AI at work?
Use AI to speed up research, analyse results, explore ideas and prepare first drafts. Check the facts, apply your own expertise and refine the work for its intended audience.
Is it safe to use ChatGPT for work?
It can be. Safety depends on the account you're using, its data settings and your organisation's requirements. Use approved platforms, and don't assume a paid subscription automatically makes all information safe to share.
Is the free version of ChatGPT OK for work?
Free and personal plans are built for individuals. If you use one for work, check whether your chats can be used for training and consider switching this off. For regular work use, a business plan gives your company more control over its data.
Is ChatGPT safe for confidential information?
It depends on your plan, settings and confidentiality obligations. Check the data terms for your account and any client agreements before sharing confidential material. Where sharing is permitted, include only what's necessary.
What information should you avoid putting into AI?
Never enter passwords, API keys, recovery codes or banking details. Beyond that, leave out personal or confidential information that isn't needed to complete the task.
How do you use AI responsibly at work?
Understand your account settings, avoid sharing unnecessary confidential information and regularly review connected apps and memory. Check accuracy, context and relevance before relying on or sharing the results.
Does an Australian business need an AI policy?
Australian businesses aren't generally required to have a standalone AI policy, but relevant privacy and confidentiality obligations still apply. A simple internal guide covering approved tools, restricted information, human review and accountability is a solid starting point.
This article is general information to help you think about how you use AI. It isn’t legal or privacy advice for your business.
