Sunday, March 30, 2025
How do people in your organization know what is acceptable to do with AI (ChatGPT, Claude, etc) and what they should avoid? While training is obviously important – what are you actually training on? Yes, how to use the tools, but how about addressing risk, and how NOT to use them?
Our AI Resource Kit includes an Employee AI Acceptable Use Policy template, which includes the following 10 elements:
Maintain Confidentiality: Do not embed into prompts anything that is either confidential to the organization, or that the organization has a duty to keep confidential (eg client or partner details, even if anonymized).
Protect Intellectual Property: Do not embed any of the organization’s intellectual property into your prompts. This can be used without permission by the system as training data for future versions of itself.
Productivity: The risk with new technologies (surfing the internet, reading social media, and now AI tools) is that it is too easy to get distracted with non-productive use, all under the heading of research. If you are using these tools, set a specific goal first; once achieved, move on.
Transparency: When using the tools, ensure that all internal stakeholders know that the AI tool was used to create the content.
Accuracy: To counter the risk of hallucinations (eg made-up assertions), all output must be fact-checked.
Overdependency: AI tools are not built for making judgements, innovating, having empathy, or drive. Yet because these tools do so well at other tasks, it’s easy to assume that they can also do well on these. They can’t, and therefore AI should only be used to inform your analysis and your decisions – not be a substitute.
Access: All access to AI tools must be done using computers on the company’s network, not using your personal devices. This will ensure that all traffic to and from the tool is protected by a firewall and other technologies.
Permission for Use: Before spending time using AI tools, you will inform your manager of the nature of your use, the expected benefits, potential downsides, and have their agreement before continuing. Agreements with your manager do not supercede this policy.
Common Sense: AI capability is changing rapidly; just because a practice isn’t in this policy, doesn’t mean that the tool should be used.
Personal Responsibility: While AI may be able to improve productivity, it has blind spots: flawed conclusions, incomplete answers, bias, and more. If you use an AI system, you remain personally responsible for all of its output.
This week’s action item:
If you don’t yet have an employee acceptable use policy, then start with this list. If you haven’t formally introduced AI in your workplace, or don’t have employee AI training, or don’t have leadership that is up-to-speed, then start with a conversation with us. (My calendar: www.RandallCraig.net/60.)
AI Insight:
While AI itself is changing on a daily basis, AI risks are not. Check out our Resource Kit. Beyond the Acceptable Use Policy template, it also includes modules, on Visioning, Training Data, Lost Opportunities, Prompt Strategy, and a QA Checklist. More at https://www.braintrustprofessionalinstitute.com/courses/ai
Does this topic resonate? Reach out to Randall: he can present it to your group. (More presentation topics)
Download Randall’s professional credentials: Speaker credentials one-sheet or Management Advisory credentials.
Content Authenticity Statement: 100% original content: no AI was used in creating this content.
@RandallCraig (Follow me for daily insights)
www.RandallCraig.com: Professional credentials site.