Check 1
Which details make an image prompt more useful? Choose all correct answers.
Useful prompts describe the subject, setting, style, purpose, point of view, and key details.
Lesson 3: Verify Before You Trust: Auditing AI Media
Step 3.11 of 6
Step 3.1 · Start here
AI can create realistic images, audio, video, graphics, logos, and designs. But realistic media may still be wrong, unsafe, fake, or not suitable to use.
In this lesson, you will study workplace-style images. You will see how prompts and training patterns shape AI images and why detailed prompts do not promise safe or correct results. You will use Pause, Trace, Check, Decide, Disclose to review made-up workplace media. Then you will choose Approve, Revise, or Reject and support your choice with evidence.
💡 Big Question and Core Rule
Big Question: How can I check media made or changed by AI? How can I decide whether it is safe and fair to use or share?
Main Rule: Do not decide based only on whether media looks real or AI-generated. Trace its , check its accuracy and context, evaluate its impact, and then decide whether it is responsible to use or share.
🧠 Memory Phrase: Pause. Trace. Check. Decide. Disclose.
🛡️ Use the Responsible AI Routine
Use this pathway whenever AI may support a task.
When to use this routine
Use it before AI supports a school, workplace, or personal task, especially when information, safety, fairness, or other people may be affected.
Your next move
Study the five steps in the image. Choose one step you think is easiest to forget and explain why it matters.
Goal → Protect → Use → Check → Own
- Protect: What private, copyrighted, sensitive, or identifying information must be protected?
- Use: What purpose, audience, and limits should guide the media?
- Check: What claims, sources, safety requirements, technical details, and professional standards must be verified?
- Own: Who is responsible for approving, revising, rejecting, or sharing the final media?
🎯 Learning Goals
By the end of this lesson, students can:
- Explain how prompts, training patterns, and model limitations shape AI-generated images.
- Explain how a media file's source differs from whether it is real, correct, or shown in context. Use Pause, Trace, Check, Decide, Disclose to find the proof you need.
- Name a needed fix, safety step, source credit, AI-use statement, or human approval.
- Check who is shown, who may be affected, and what safety or disclosure steps are needed. Use evidence to choose Approve, Revise, or Reject.
✅ You will show success by: explaining how prompts and patterns shape AI media, distinguishing from accuracy and responsible use, identifying credible evidence, evaluating community effects, recommending a safeguard, and justifying an Approve, Revise, or Reject decision.
⏱️ Pacing at a Glance
| Lesson Step | Time |
|---|---|
| 3.2 Connect & Learn: Creating and Questioning AI-Generated Images | 10 minutes |
| 3.3 Learn: How AI Image Generators Create Images | 6 minutes |
| 3.4 Learn: Verify | 8 minutes |
| 3.5 Apply Your Learning: AI Media Audit Lab | 14 minutes |
| 3.6 Show What You Know: AI Media Creation and | 12 minutes |
| Total | 50 minutes |
📚 Key Words to Know
AI Image Generation & Prompts
: A tool that creates images from text prompts by using patterns learned from .
: A written description that tells an what to create.
: Revising a by adding, removing, or changing details to improve the result.
: Examples used to help an AI system learn patterns between words and visual features.
: A type of image model that begins with random noise and gradually changes it into an image.
, Media & Ethics
: Images, video, or audio created or substantially changed by AI.
: AI-created or edited media that makes a real person appear to say or do something that did not happen.
: False or wrong information that may mislead people. (Disinformation is false information shared on purpose to deceive).
: Information that shows where media came from, who created it, and how it was changed.
: Checking media by using its original source, trusted evidence, official information, workplace rules, or qualified experts.
Step 3.2 · Learn and check
Connect & Learn: Creating and Questioning AI-Generated Images
Estimated time: 10 minutes
Responsible AI Routine: Goal → Protect → Use → Check → Own
🔎 Connect & Learn: Creating and Questioning AI-Generated Images
Review Images A–D. Select one image you would pause to evaluate before using in a workplace context.
- Image A: Hand with unusual finger structure and wrist tattoo.
- Image B: Person in a store aisle holding a bag.
- Image C: Outdoor site workers near tools and buildings.
- Image D: Welder at a metal table with bright sparks.
💭 Self-Reflection Questions
As you review the images, consider the following questions to guide your analysis:
- Which image did you choose, and what specific visual detail made you pause?
- What additional evidence or context would you check before sharing or publishing this media?
- Who could be affected if an inaccurate, biased, or misleading image is used in a professional setting?
Could You Tell for Certain?
Some AI-generated images have obvious mistakes, while others look polished. Real photos can also be blurry, edited, cropped, poorly lit, or paired with a misleading caption. Looks alone cannot prove that media is real, accurate, or responsible to use.
A visual clue can begin an investigation, but it is not proof. Reviewers trace where the media came from, check its claims and context, and consider who may be affected. Then they decide whether to approve, revise, reject, or label the media.
Pause → Trace → Check → Decide → Disclose
▶️ Your next move: Watch for how details in a change an AI-generated image.
This video may display YouTube ads.
📖 Your next move: Read the short excerpt below. Focus on how AI image prompts work.
An AI image tells the tool what you want it to create. A strong may name the main subject, background, colors, style, mood, lighting, and point of view. The tool uses these details to create visual features that are likely to fit the request.
The first image may not match your goal. Improve the by adding or removing details. You can change the angle, lighting, mood, or focus. Small changes may create large differences. Treat each result as a draft and keep revising until it fits the purpose.
Check 2
Put the image-refinement process in the order that best supports a focused, evidence-based revision.
Choose the steps in order, from first to last.
Strong refinement begins with a clear goal, uses evidence to target one important mismatch, and checks whether the revision improved the result.
Check 3
A reviewer notices an unusual hand in an image. What is the most responsible conclusion?
A visual problem can start an investigation, but it does not prove the image’s origin or accuracy.
Step 3.3 · Learn and check
Learn: How AI Image Generators Create Images
Learn: How AI Image Generators Create Images (6 minutes)
Estimated time: 6 minutes
Responsible AI Routine:
Goal → Protect → Use → Check → Own
▶️ Your next move: Watch for how training examples help an image generator learn patterns.
📖 Your next move: Continue with the short reading and examples below.
An creates images from text prompts. During training, it learns patterns that connect words with visual features such as objects, colors, styles, lighting, mood, and camera angle. The system does not imagine or understand the scene like a person. It creates features that are likely to match the .
The result depends on the , the system’s design, and the . Missing or inaccurate patterns may cause mistakes or stereotypes. A detailed may improve the image, but it cannot guarantee readable words, correct measurements, safe steps, fair representation, original work, or permission to use the result.
Source Foundation
Michigan Virtual, “How Do AI Image Generators Work?” — A student-friendly explanation of training patterns and image generation. Source video
Ho, Jain, and Abbeel, “Denoising Diffusion Probabilistic Models” — The foundational research paper describing diffusion-based image generation through a progressive denoising process. View the diffusion-model research paper
NIST, Generative Artificial Intelligence Profile — Describes how , model design, and generative-system risks can affect output reliability and representation. View the NIST Generative AI Profile
Last Updated: August 4, 2026
Check 1
How does an AI image generator create features that fit a text prompt?
An image generator uses learned statistical patterns to create visual features that fit the prompt.
Check 2
Which statements describe limits of AI image generators? Choose all correct answers.
Image generators may create convincing errors and repeat patterns or stereotypes from their training data.
Check 3
Why can a detailed image prompt still lead to wrong or unsafe details?
Prompt detail can improve relevance, but it cannot remove limits in the model, data, or design.
Step 3.4 · Learn and check
Learn: Verify
Learn: Verify
Estimated time: 8 minutes
Responsible AI Routine: Goal → Protect → Use → Check → Own
▶️ Your next move: Watch for one way to verify an image when visual clues are not enough.
📖 Your next move: Read the three evaluation questions and 5-step process below.
3 Key Questions Before Trusting Media:
- & Authenticity: Where did it come from, who created it, and how was it changed?
- Accuracy & Context: Are the image, caption, event, measurements, and claims correct?
- Responsible Use & Impact: Is it fair, safe, allowed, right for the audience, and clearly labeled?
The 5-Step Media Check:
- Pause: Do not share or approve media only because it looks convincing.
- Trace: Find the original source, creator, caption, date, context, credits, and edit history/Content Credentials.
- Check: Compare claims with trusted sources, workplace rules, specifications, and experts. Verify safety, fairness, privacy, and permissions.
- Decide: Choose Approve, Revise, or Reject based on evidence and potential impact.
- Disclose: Clearly explain or label AI use when required by policy or platform rules.
Note: An AI detector score or visual flaw is a clue, not definitive proof. Neither should be used as the sole evidence for a decision.
Source Foundation
- Michigan Virtual, “Recognizing AI-Generated Images”
- NIST, Reducing Risks Posed by Synthetic Content
- C2PA, Content Credentials Explainer
Last Updated: August 4, 2026
Check 1
A reviewer notices a warped logo and an inconsistent shadow. What should happen next?
Visible flaws can raise questions, but reviewers still need source, context, and evidence checks.
Check 2
Which information belongs in the Trace step? Choose all correct answers.
Tracing looks for the original source, creator, date, caption, context, credits, and available history.
Check 3
A detector says an image is 92% likely to be AI-generated. What is the strongest next step?
Detector scores can support an investigation, but source, context, and factual claims still require checking.
Step 3.5 · Practical application
Apply Your Learning: AI Media Audit Lab
Time: About 15 minutes
What you will do: Audit one image pair by answering four guided questions.
What you need: This activity page. No outside AI detector is needed.
What you will submit: Follow your instructor's directions for the completion reflection.
Start here: Choose the program that best fits you and select one image pair.
is not automatically safe or appropriate to use. In this activity, you will examine a workplace image and decide whether it is ready to use or needs further review.
For your selected comparison:
- Examine the image carefully.
- Identify which image has an obvious concern.
- Identify any concerns.
- Choose the strongest review decision and the evidence or expert to check next.
- Review the feedback before continuing.
Important note: You are evaluating accuracy, safety, permission, privacy, and context — not trying to determine whether AI created an image. A polished image still needs normal before approval.
Fictional practice materials: All comparison examples in this activity are fictional. Some intentionally include unsafe, inaccurate, misleading, private, or incomplete content so you can practice identifying problems. Do not treat them as real records or approved guidance.
Interactive image comparison
Career Program Media Audit
Choose your program
Study one image pair from your career program. Four questions will guide you through the clues, concern, review decision, and responsible next step.
Finish this comparison before choosing another program for optional practice.
Welding Media Audit
|
Image A |
Image B |
Image A needs review. Safety glasses alone do not provide the required protection for the welding shown. Safety materials must model the correct helmet and complete PPE. Main concern: It models incomplete eye and face protection while welding. Responsible next step: Replace or revise the image so it demonstrates required welding PPE. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Agriscience and Veterinary Medicine Media Audit
|
Image A |
Image B |
Image B needs review. Home remedies and dosage-like directions can harm an animal or delay needed care. Veterinary guidance requires an approved source or qualified professional. Main concern: It gives unverified animal-health treatments and dosage-like directions. Responsible next step: Have a veterinarian or approved source verify every care recommendation. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Public Safety Media Audit
|
Image A |
Image B |
Image B needs review. A safety or recruitment visual should model complete PPE. Shorts and athletic shoes are not appropriate protection for the live-fire scene shown. Main concern: It shows incomplete protective equipment during live-fire training. Responsible next step: Replace or revise the image so it models required protective equipment. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Mechatronics Media Audit
|
Image A |
Image B |
Image B needs review. Wiring, setup, and operating details can damage equipment or hurt people when they are wrong. Verify the guide with approved schematics, manuals, and shop procedures. Main concern: It presents equipment setup and safety guidance that needs technical . Responsible next step: Verify the schematic and safety steps before testing the equipment. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Marketing and Graphic Design Media Audit
|
Image A |
Image B |
Image A needs review. A polished logo can still conflict with an existing brand. Check trademarks and originality before using an AI-generated identity for a business or client. Main concern: Its name and logo require a trademark and originality review. Responsible next step: Run a trademark check and revise the identity if it is too similar. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
U.S. Army JROTC Media Audit
|
Image A |
Image B |
Image B needs review. Fitness guidance should include safe progression, recovery, supervision, and attention to pain or injury. Motivation does not override health and safety. Main concern: It promotes unsafe exercise expectations and dismisses signs of injury. Responsible next step: Replace it with approved, age-appropriate fitness and safety guidance. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Health Science and Medical Careers Media Audit
|
Image A |
Image B |
Image A needs review. Patient-facing health guidance must be accurate, current, and appropriate for the audience. A disclaimer does not replace by an approved medical source or licensed professional. Main concern: It gives health guidance that could be misunderstood as patient-specific advice. Responsible next step: Have an approved medical source or licensed professional verify the handout. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Forensic Science Media Audit
|
Image A |
Image B |
Image A needs review. A realistic mock case can be mistaken for authentic evidence, and identifying details create privacy concerns. Label simulations clearly and use fictional, non-identifying information. Main concern: It could be mistaken for real evidence and exposes identifying details. Responsible next step: Label it fictional, remove identifying details, and obtain teacher approval. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Education Media Audit
|
Image A |
Image B |
Image B needs review. Classroom materials should not reinforce stereotypes or teach incorrect content. Review for accuracy, representation, age appropriateness, and accessibility. Main concern: It includes stereotypes and instructional information that may be inaccurate. Responsible next step: Correct the content and review it for accuracy, bias, and age fit. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Culinary Arts Media Audit
|
Image A |
Image B |
Image A needs review. An absolute allergy-safe claim can cause serious harm if ingredients or cross-contact controls are wrong. Verify actual recipes, labels, and kitchen procedures. Main concern: It makes an absolute allergen-safety guarantee without . Responsible next step: Verify ingredients and cross-contact controls, then revise the claims. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Construction Trades Media Audit
|
Image A |
Image B |
Image B needs review. Construction and temporary-wiring guidance can create real hazards when details are wrong. Check current plans, code, manufacturer information, and instructor directions. Main concern: It presents technical building guidance that may be unsafe or incorrect. Responsible next step: Check the plan, code, and instructor directions before correcting the guide. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Computer Science Media Audit
|
Image A |
Image B |
Image A needs review. Recognizable company logos and placement claims can imply endorsement. Verify permissions and evidence for claims before publishing. Main concern: It may imply endorsements and use brand assets without permission. Responsible next step: Verify each claim and permission, then remove unsupported brand references. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Cosmetology Media Audit
|
Image A |
Image B |
Image B needs review. Claims such as no patch test and no damage can put clients at risk. Chemical-service materials must follow product directions, allergy precautions, and professional safety rules. Main concern: It makes unsafe and unsupported chemical-service claims. Responsible next step: Check product directions and safety rules, then remove unsafe claims. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
Automotive and Medium/Heavy Diesel Media Audit
|
Image A |
Image B |
Image A needs review. Supporting a caliper by its hose and lubricating a rotor face are unsafe instructions. Repair guidance must match approved service information and shop procedures. Main concern: It shows unsafe and incorrect brake-service steps. Responsible next step: Verify the procedure with service information and correct the unsafe steps. Image B has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
3.5 Practice: AI Media Audit
Review both images for each program area. Choose the assessment and next step that best fits the pair. Focus on responsible use, not on guessing whether AI created an image.
Aviation Media Audit
|
Image A |
Image B |
Image B needs review. It presents unsafe flight advice as simple guidance. Operational aviation information must come from approved sources and established procedures. Main concern: It gives unsafe and inaccurate flight guidance. Responsible next step: Replace the guidance with procedures from approved aviation sources. Image A has no obvious visual concern, but appearance alone does not prove that it is ready. A reviewer must still check its source, claims, permissions, context, and intended use.
You completed your program comparison. You may stop here or practice with another program.
This is a practice activity and does not need to be submitted.
Step 3.6 · Lesson assessment
Show What You Know: AI Media Creation and
Your place in this lesson is saved automatically on this device.