Free preview. This lesson is part of a paid course that earns the ETA Specialist: AI in Manufacturing credential. About this course
Ground rules: safety, data and control systems
In this lesson you'll
- Name the ground rules for AI in manufacturing
- Protect proprietary process data, customer data and export-controlled technical data
- Explain why AI tools stay out of plant control systems unless OT security approves
Welcome
Welcome to AI in Manufacturing from the Enterprise Technology Association (ETA). Manufacturers are using AI for maintenance, quality, planning and training, often starting in the office and moving toward the floor. This course is about getting real value while protecting the things that matter most in a plant: people's safety, product quality, intellectual property and uptime. It assumes you've completed Level 1: Foundations of the ETA Certified AI Empowered Professional.
- Step 1: Assess
Is this a good task for AI, and what's at stake?
- Step 2: Direct
Brief the AI clearly: goal, context, constraints, done criteria.
- Step 3: Assure
Check the output before you rely on it.
- Step 4: Protect
Keep people, data and your organization safe.
- Step 5: Tune
Iterate, then save what works for next time.
Use the steps in order on every task, and loop back to Tune when results fall short.
Four ground rules
- Safety decisions stay with qualified people. AI can help find information. It doesn't decide whether equipment is safe to run, whether a guard can come off, or how to lock out a machine.
- Data stays where it's allowed. Process recipes, customer drawings, pricing and controlled technical data go only into approved tools, if at all.
- Control systems stay isolated. AI tools don't connect to PLCs, SCADA or other operational technology (OT) unless your OT security team designs and approves it.
- Numbers get checked. Torque values, tolerances, setpoints, chemical quantities and dates are verified against the controlled source before anyone uses them.
Data rules
Passwords, access codes, security keys and other credentials
- Personal information
Social Security numbers, birth dates, home addresses, personal phone numbers
- Health information
Diagnoses, medical leave details, patient records
- Financial account details
Bank and card numbers, individual pay records
- Student records
Grades, discipline and other student information
- Confidential business and client information
Unreleased results, pricing, contracts, legal matters, anything under a confidentiality agreement
Some organizations approve specific tools for some of this data. That decision belongs to your organization, not to you. If you're unsure, treat it as off-limits and ask. You rarely need real details: redact, substitute or generalize.
| Usually OK in an approved tool | Ask first | Never |
|---|---|---|
| Public manuals and standards, your own non-sensitive procedures, anonymized production data | Customer drawings and specifications, supplier pricing, process parameters, quality data tied to a customer | Export-controlled technical data (ITAR, EAR) unless trade compliance approves the tool, trade secrets in unapproved tools, employee personal data |
Export controls. If your plant makes defense articles or controlled items, technical data about them may be export controlled under ITAR or the EAR. Sharing it with a tool whose servers or staff are outside approved locations can be an export. Your trade compliance lead decides which tools, if any, can handle controlled data.
Customer obligations. Contracts and NDAs often restrict where customer drawings and specifications can go. Check before uploading.
Operational technology
Plant networks and control systems are built for safety and uptime, often with equipment that can't be patched quickly. Connecting a new AI service to them can open security holes. Keep AI tools on the business network, using exported data, unless your OT security team, following guidance such as NIST SP 800-82, approves a specific design.
Workplace example: a precision machining job shop
An engineer wanted AI help writing a setup sheet and uploaded a customer's drawing for a defense subcontract into a public AI tool. The quality manager caught it during a document review and reported it to the owner, who called their trade compliance consultant. The shop then set a rule, posted at every engineering workstation: customer drawings only in the approved tool, and never for controlled parts.
Try it
Write your plant's AI ground rules in your approved AI tool, then check them yourself:
Goal: Draft one-page AI ground rules for our plant.
Context: We make [products] for [customers/industries]. We use [AI tools].
Audience: Engineers, supervisors and office staff.
Constraints: Cover safety decisions, data rules (OK / ask first / never), export-controlled data, control systems and checking numbers. Mark anything uncertain "confirm with [EHS / trade compliance / IT]."
Format: Four headings, 3 to 5 bullets each.
Done criteria: Every rule is something a person can check on the floor.Review the draft with your EHS, IT and trade compliance contacts.
Key takeaways
- Safety decisions stay with qualified people; numbers get checked against controlled sources.
- Process, customer and export-controlled data go only where approved; trade compliance decides on controlled data.
- AI tools stay off control systems unless OT security designs and approves the connection.
Sources