A client asked me a question earlier this year that I wasn’t ready for. We were sitting in a strategy meeting talking about content, and out of nowhere she said, “So what’s our AI policy?”

I gave her a good answer. But I remember thinking afterwards that I didn’t have a great answer written down. And if I didn’t, most of the businesses I know probably didn’t either.

That conversation is why Lifedge has one now. And it’s why I want to walk through what should be in yours.

The Quiet Truth About AI Policies

Here’s the honest version. Whether you’ve written one down or not, your business already has an AI policy. Your team is already making decisions about it every day.

They’re deciding whether to draft a client email with ChatGPT or write it themselves, whether to upload a private document into an AI tool to summarize it, whether to let AI generate images for a customer-facing campaign, and whether to disclose that AI was used at all.

If those decisions aren’t guided by something written down, they’re being made in a dozen different ways across your team. Some of them are probably fine. Some of them are probably not. And you won’t know which is which until it becomes a problem.

An AI policy isn’t a document you write to be trendy. It’s a document you write so everyone on your team knows what “yes” and “no” look like without having to ask you every time.

How Lifedge Built Ours

When we sat down to write ours, we didn’t try to solve every future AI question in one document. We started with the questions we were actually facing.

Four questions turned out to be the core.

1. What Can We Use AI For, and What Can’t We?

We wanted to be specific. Not “use AI responsibly,” because that means nothing. So we listed the kinds of tasks AI is welcome in (research, first drafts, brainstorming, summarizing internal notes, generating structural options) and the kinds it’s not (final client-facing copy without human review, confidential client data upload into free-tier tools, generating images of real people, replacing a strategic conversation).

2. What Client Information Can Go Into Which Tools?

This one matters most. Some AI tools train on the data you feed them. Others don’t. Some are contractually safe for confidential information. Others absolutely are not. Our policy names specific tools that are approved for which kinds of information, and one hard rule: if you’re not sure, ask before you paste.

3. When Do We Disclose That AI Was Involved?

The short version of our answer: when a client would reasonably want to know, yes. When it’s the same category of tool we’d use to spellcheck or format, no. The gray zone is real, and we err on the side of transparency in it.

4. Who’s Responsible for the Final Work?

This one closes the loop. AI doesn’t own our work. The person on our team who put it in front of a client does. That means every AI-assisted piece is read, checked, and signed off by a human before it goes out. Not because AI is untrustworthy, but because responsibility can’t be delegated to a tool.

How We Actually Use AI at Lifedge

A policy without practice is just paper. Here’s what our AI use actually looks like day to day.

We use AI to prepare for client meetings. Before each strategy call, we pull together a dossier that combines meeting notes, task history, and analytics into a single brief. AI helps us stitch that together in minutes instead of hours. The thinking is still ours. The pattern recognition is still ours. But the assembly is faster.

We use AI to brainstorm structural options. Before we write a homepage, we’ll often have AI generate five different structural approaches for the same brief. Not to use verbatim, but to give us more shots on goal than a single human draft would.

We use AI to catch things a human would miss. Consistency across a long document. Broken internal links. Places where our tone slipped from confident to hedging. AI is good at the tedious pattern-matching that used to eat our afternoons.

We don’t use AI to replace the strategic conversation with a client, to generate final client-facing copy without a human writer touching it, or to make judgment calls about someone’s brand voice, mission, or audience. Those decisions belong to humans who are accountable for them.

A Starter Framework You Can Use

If you don’t have a policy yet, here’s a starting point. Sit with your team and answer these questions on paper.

1. What are the top three tasks your team is already using AI for, informally? Name them.
2. What kinds of information should never go into a public AI tool at your company? Be specific.
3. When a client asks whether AI was involved in the work you did for them, what do you want your team to say?
4. Which one person on your team is responsible for reviewing AI-generated work before it leaves the building?

Those four answers are 80 percent of a working AI policy. Everything else is refinement.

The Bigger Picture

An AI policy is not really about AI. It’s about the kind of company you want to be in an era where the tools are changing faster than most policies can keep up.

The businesses that will earn the most trust in the next few years are the ones that were early to be clear: clear with their team about what’s allowed, clear with their clients about what’s happening, and clear with themselves about who’s accountable for the work.

You don’t need a legal document. You need a conversation, written down.

If you’d like a copy of our AI policy as a starting point for yours, reply to this post or send us a note. We’re happy to share what we’ve learned and think it through with you.