Let employees use AI through an approved process: define which tools and work tasks are allowed, what information may be entered, and who reviews the risks. Keep client, patient and other sensitive data out unless the specific service and workflow have been assessed and authorized. Assign owners to check access, retention, integrations and incident response before rollout.
In This Article
- What should an Atlanta business decide before allowing AI use?
- Which information should employees keep out of AI prompts?
- What should IT and compliance check about an AI service?
- How can employees follow AI rules in daily work?
- How should an Atlanta business review its AI controls?
- Frequently Asked Questions
What should an Atlanta business decide before allowing AI use?

“John brought the right knowledge to my issue and resolved it in about a reasonable amount of time. I walked away confident the problem was actually fixed.”
Start with the work employees want AI to help with. Drafting a generic meeting agenda is different from summarizing a patient record, reviewing a client contract or analyzing a tax return. Identify the information involved and the consequences of an inaccurate or exposed result before approving the use.
Write down which tools and accounts are approved, what information employees may enter, and which tasks are prohibited or require review. Include browser-based tools, extensions, transcription services and AI features built into software already in use. A familiar application can include a new AI feature with separate data handling or access settings.
For each approved use, identify a business owner and a technical owner. The business owner confirms that the task is appropriate and that a person remains accountable for the result. The technical owner checks configuration, access and data handling. Involve legal or compliance advisers when the organization needs an assessment of its specific obligations.
Set four boundaries for each approved use: tool, task, permitted data and accountable owner.
Those boundaries make the decision usable in daily work. A broad instruction to "use AI responsibly" does not tell an employee whether a particular document, prompt or connected application is allowed.
Keep an approval record with the intended purpose, data categories, configuration and review date. Update it when a workflow changes, rather than assuming an earlier review covers a new feature or a different kind of information.
The goal is not to approve AI in the abstract. It is to decide which work can be supported safely and under what conditions.
Which information should employees keep out of AI prompts?
Unless a specific tool and workflow have been reviewed and approved, tell staff not to paste information that identifies a client, patient, employee or another individual. Examples include names, contact details, account identifiers, medical information, credentials, financial records, legal documents and confidential business information.
Removing a name may not be enough. A combination of dates, locations, unusual facts or case details can still identify someone. For testing prompts or learning a feature, use fictional examples or information minimized so the task does not require identifiable or confidential details.
Give staff examples relevant to their roles. A medical office can distinguish a generic patient-education draft from a prompt containing a patient's clinical history. A law firm can separate a general outline from a client's privileged facts. A CPA firm can use a fictional return scenario instead of a client's tax records.
For businesses handling protected health information, review whether the particular service and intended use meet the organization's HIPAA obligations before entering that information. GDS Technology provides IT compliance services, including HIPAA-related services, but technology work does not replace the organization's own legal and compliance decisions. See the HIPAA IT compliance service overview.
Apply data minimization by default: give a tool only what the approved task requires. If the task can be completed with a generic description or fictional example, do not include a real person's record.
Make the boundary easy to remember and easy to check. Employees should know what to do when they are unsure, rather than making a high-stakes judgment about sensitive data on their own.
Keep identifiable or regulated information out unless the exact workflow has been explicitly reviewed and cleared.
What should IT and compliance check about an AI service?
Review the service's actual terms and administrative controls, not just its product description. Confirm how prompts and uploaded files are handled, whether they may be used to improve models, how long data is retained, and what deletion options are available. Check whether these terms or settings vary by account type or feature.
Establish who can use the service and how access is removed when a person changes roles or leaves. Prefer organization-managed identities and accounts where available. Determine whether administrators can manage users, settings, audit information and connected applications. Avoid putting business work into personal accounts the organization cannot govern.
Check how information moves into and out of the service. Integrations, browser extensions, connectors, file uploads and automated workflows may expose more information than a prompt alone. Approve only connections needed for the use case, confirm what they can access, and review them when the workflow changes.
For Atlanta organizations, Georgia's data-breach notification law is a reason to include incident handling in the review. Decide who evaluates a suspected disclosure, preserves relevant records and coordinates with counsel. Do not assume a vendor setting or a general AI policy settles the organization's legal obligations.
Record the decision, approved configuration and person responsible for review. GDS Technology's IT compliance services include scoped planning and evidence coordination around technology controls. That work supports a practical improvement plan; it is not a legal opinion or a certification.
Vendor review and access control make policy actionable. A written rule without ownership, configuration and a review process is difficult to apply consistently.
How can employees follow AI rules in daily work?
Publish a short, role-based standard employees can consult at the point of work. Name approved tools, permitted tasks, restricted data, review requirements and a contact for questions. Explain how to report an accidental disclosure promptly, without asking employees to investigate it themselves.
Train with realistic examples from the organization's work. Ask employees to identify which prompt is safe, what information should be removed and when a person must verify an output. For high-impact decisions, require an accountable person to check source material, correct errors and decide whether the result is appropriate to use.
Provide an exception path. An employee who sees a useful but unapproved task should be able to request review rather than quietly using a personal account. Record the purpose, data involved, business owner and safeguards considered. Then approve, revise or decline the request and explain the decision.
When an employee encounters a new AI use case, make the next steps straightforward:
- Describe the work task and the information the tool would receive.
- Ask the designated business and technical owners to review the tool, data handling and safeguards.
- Wait for the decision, then use only the approved workflow and follow its human-review requirements.
One professional-services customer described the value of knowledgeable help and a clear resolution: "John brought the right knowledge to my issue and resolved it in about a reasonable amount of time. I walked away confident the problem was actually fixed." Madhav Naik, professional services. In an AI governance process, the same practical principle matters: give people a clear route to an answer, then check that the control works.
Keep the guidance close to the work. A policy is more useful when employees can apply it during a busy day, recognize a situation that needs approval and report a mistake without delay.
Clear instructions and a workable exception path help employees use approved tools without improvising around the rules.
How should an Atlanta business review its AI controls?
Assign an owner to revisit approved tools, use cases and settings when a vendor changes a feature, the business adds an integration or the information involved changes. Review access periodically and remove accounts and connections no longer needed.
Include AI use in incident and continuity planning. Decide how staff report a suspected disclosure, who assesses the event and how the business preserves relevant information. If a workflow supports a medical office or another operationally sensitive service, identify the work that depends on it and the fallback if access is suspended.
Keep useful evidence: the current approved-tool list, completed training, exception decisions, access reviews and incidents or near misses. This gives management a way to see whether employees can follow the process and where clearer guidance or a configuration change is needed.
GDS Technology brings technology, cybersecurity and compliance planning together around business operations. Its cybersecurity services and managed IT services can support organizations reviewing technology controls and operational needs. Start by mapping the information, people and systems involved, then decide which use cases can be governed clearly enough to support the work.
Review the process when the business, its tools or its data changes. Regular ownership keeps AI use aligned with operational needs and the organization's responsibilities.
More on this from Built Wired Secured - Blog: How to Label Data Cabling Without Creating Risk — Built, Wired & Secured.
Frequently Asked Questions
Can employees use public AI tools for work?
Employees can use a tool for work only when the organization has approved that tool and the specific task. Set rules for information they may enter, account ownership, retention, integrations and human review. Keep client, patient and other sensitive information out unless the exact service and workflow have been reviewed and authorized by the organization.
Can a business use AI with HIPAA-regulated information?
Do not enter protected health information into an AI service until the organization has reviewed the service, configuration, intended use and applicable HIPAA obligations. Confirm data handling and access terms, and involve the organization's compliance and legal advisers as needed. Technology safeguards support compliance work but do not establish compliance by themselves.
What should an employee do after pasting sensitive information into an AI tool?
Report it promptly through the organization's incident process. The responsible team should identify what information was entered, which account and service were used, and whether the data can be removed or access restricted. Preserve relevant details and involve the appropriate privacy, security and legal owners to assess notification or other obligations.
How do we stop staff from using unapproved AI tools?
Offer an approved route that meets legitimate work needs, explain the boundaries with role-specific examples and provide a simple way to request a new use case. Review account access, browser extensions and integrations within the organization's governance capabilities. Pair technical controls with training and a clear incident-reporting path.
Follow GDS Technology on Facebook, and subscribe on YouTube for new videos.