Write a short note that tells colleagues how they may use AI-generated text and images.
When a team starts using AI to make text and images, colleagues tend to ask the same few things: which tool they can use, what to check before posting, and whether to tell anyone that a picture was made by AI. If the answers live only in one person's head, each new hire gets a slightly different version of them, and nobody gets an answer when that person is away.
A content use note fixes this. It is one page that tells colleagues how AI-generated text and images may be used in the team's work. It is narrower than the full team AI policy you will write in module 8, and you can fold it into that policy later. For now it only needs to cover content.
A useful note answers four questions, each in a few short sentences.
The first is tools and accounts. Name the tools allowed for content, for example a specific image generator or the company's approved assistant, and say which account or plan. The account matters, and lesson 1.2, "Personal, business and enterprise accounts are not the same", explains why. If something is not allowed, such as using personal accounts for client work, the note should say so.
The second is what gets checked before publishing, and who checks it. Lesson 4.3, "Working rules for AI content at work", gives you the checks to choose from. Say which content needs a reverse image search or a plagiarism check, and name the role that signs off. On a small team, sign-off might sit with the person publishing. A larger team might need a second person to check anything paid or printed.
The third is when to disclose AI use. This is a judgement call with no single right answer. Some teams disclose on any content that is mostly AI-generated. Some disclose only when a client's contract requires it. Others disclose only when a reasonable reader would care, for example a photo that looks like a real event. Check whether your clients, your industry or the platforms you post on have their own disclosure rules. Whichever rule you choose, write it down so everyone does the same thing.
The fourth is complaints. Say who receives a complaint, that the content comes down while it is looked into, and where the record of how the content was made is kept. Lesson 4.2 explains why a fast, calm response matters more than winning an argument.
Kelvin runs a bakery. His cousin joined last month to help with marketing, and on her second day she asked him three things: whether she could use the image generator for Instagram, what she needed to check before posting, and whether they should tell customers when a picture was AI-made. Kelvin answered each question out loud. Afterwards he realised the next new hire would ask the same questions, and he might not be there to answer them.
He wrote a note titled "How we use AI for bakery content". The details below are his choices for his bakery, so treat them as examples of the kind of thing to write rather than rules to copy.
Tools: the image generator on the bakery's paid business account, and the approved assistant on the shared work login. No personal accounts for anything posted under the bakery's name. Social posts: the person posting reads the text out loud and checks every fact, price and date before it goes up. Anything printed or paid for, such as packaging, flyers and ads: run a reverse image search on every AI image and show it to Kelvin before it goes to the printer. Prompting: never prompt with the name of an artist, a cartoon character or another brand. Telling customers: they do not label every post, but they do disclose if an image shows a product, person or scene that looks real but is not, for example a staged shopfront photo. They never use AI images in place of real photos of their actual cakes. Records: save the prompt, the original image and the edited version in the campaign folder on the shared drive. Complaints: forward them to Kelvin the same day, take the post down while it is looked at, and do not argue in the comments.
Every line answers a question someone actually asked, and the whole note fits on one page.
Start by writing down the questions colleagues have asked, or would ask, about AI and content. If nobody has asked yet, imagine a new hire's first week and list what they would want to know before posting anything.
Then draft each of the four sections in plain language, using your team's actual tools, platforms and roles. Specific wording works better than vague wording: "the marketing lead checks paid ads" tells someone what to do, while "content is reviewed appropriately" does not. If a section runs past five or six sentences, you are probably trying to cover every possible case. Cover the common cases and say who to ask about the rest.
Next, check the note against your employer's wider policy and the tools' terms, which is rule one of lesson 4.3. Your note can be stricter than either of them, but it cannot allow something they forbid.
A finished note is one page or less. It names real tools, real accounts and real roles, and a new colleague can read it in two minutes and know what to do with the next image they generate. It has also been read by at least one person other than you. This last part matters most, because a second reader will find a gap you missed.
When Kelvin's cousin read his draft, she pointed out that it said nothing about AI-written captions in Malay and Chinese, which nobody on the team could check properly. He added a line to cover it. Write your note, then give it to a colleague to read, and expect them to find a gap like the one Kelvin's cousin found.
Write a content use note no longer than one page. Show it to one colleague and ask if it answers the questions they actually have.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).