AI-Generated Images: Ethics and Responsibilities

CreatiCode

AI-Generated Images: Ethics and Responsibilities

1. Introduction

What You Will Learn:

How AI uses existing images to create new artwork (the “remix” process).
Why copyright may be an issue when AI replicates famous styles or characters.
How AI can sometimes produce offensive, biased, or racist images.
The dangers of misinformation with AI-generated images.
Ways to think critically about using AI responsibly and ethically.

Why It Matters:
AI image-generation tools are more accessible than ever. They can help us create fun, imaginative pictures — but they also raise important questions. As you learn to use AI image generation tools on CreatiCode, remember that your choices can affect others and reflect on broader social issues.

2. How AI “Remixes” Existing Artwork

2.1 AI creates new images by remixing existing images

Most AI tools learn by looking at huge collections of existing images. Through mathematical patterns, the AI figures out shapes, colors, and styles. When you give it a prompt (like “A dog playing the piano in the style of Van Gogh”), it pulls together what it has “seen” about dogs, pianos, and Van Gogh’s painting style. It then remixes those elements to create something new.

Activity:
- Go to CreatiCode and open the AI image tool for backdrops;
- Type the prompt: “A dog playing the piano in the style of Van Gogh.” to do a search
- Look at the search results.
- Question: Can you spot any hints of Van Gogh’s style (like swirling brush strokes or bright contrasting colors)?
Reflect:
- “Do you think the AI is being creative, or is it just copying bits and pieces?”
- “How is this different from a human artist being inspired by someone else’s work?”

Takeaway: AI is powerful at combining elements it has seen before. However, the boundaries between “inspired by” and “copied from” can be blurry.

3. Copyright and Ownership

3.1 What Is Copyright?

Copyright is a legal protection that gives creators control over how their work is used. If you paint a picture, you generally have the right to say who can copy it, sell it, or display it.

3.2 Does AI Infringe on Copyright?

There is currently no clear legal consensus on whether AI-generated content constitutes copyright infringement. The issue remains highly contested, with credible arguments on both sides:

Argument for Infringement: Critics argue that if an AI system is trained on copyrighted works—particularly when those works are distinctive and recognizable — then the output may effectively constitute an unauthorized derivative. If the generated content closely resembles a specific artist’s style or reproduces identifiable elements from their work, it may cross the line from inspiration into copying.
Argument Against Infringement: On the other hand, proponents of AI-generated art contend that these systems operate much like human creators — by learning patterns, styles, and techniques from a broad array of inputs. In many cases, the final output is seen as a new, transformative work rather than a reproduction. From this view, the use of training data is akin to artistic influence rather than direct appropriation, and thus should not be considered infringement.

3.3 Famous Cartoon Characters

Let’s say you type, “Generate an image of Mickey Mouse dancing.” This is clearly using a Disney-owned character.

Questions to think about:

Is that considered using someone else’s creation without permission?
Would it matter if you were only doing it for fun, or for selling T-shirts?

Activity:

Hypothetical Scenario: You generate a digital poster featuring Mickey Mouse (or Spider-Man) to print on T-shirts you plan to sell.
Ask Yourself:
- “Am I allowed to sell T-shirts with these characters?”
- “Would it be okay if I gave them away for free instead?”
- “Is there a difference if the AI image is slightly off-model but still recognizable?”

Takeaway: The law is still catching up with AI technologies, but you should always be cautious about using recognizable characters or artworks without permission — especially when money or public distribution is involved.

3.4 How to avoid copyright issues?

Although it is highly unlikely for Disney to go after you if you create a new project using the image of Mickey Mouse on CreatiCode, it is still strongly recommended for you to avoid such issues. Here are some practical guidance for what you can do:

Refer to “style” instead of “character” in your prompt: In general, an artistic style is not copyrighted, but specific characters are copyrighted.
Refer to earlier styles that are now in the public domain: for example, an artistic style from 100 years ago is usually regarded as “public domain” as opposed to proprietary, so it is much safer to use them.

Here are some examples:

Instead of asking for “a turtle in the style of Mickey Mouse”, ask for “a happy cartoon turtle in 1930s rubber hose animation style, with bendy limbs, pie-cut eyes, and vintage gloves.”.
Instead of asking for “a blue hedgehog like Sonic”, ask for “a stylized blue animal character in 90s arcade game art style, with speed lines and neon highlights.”
Instead of asking for “a superhero like Batman”, ask for “a masked vigilante in a dark, moody noir comic style, with dramatic shadows and a gothic skyline in the background.”
Instead of asking for “a robot like Wall-E”, ask for “a small boxy robot with tank treads and binocular eyes, illustrated in a retro-futurist 1960s sci-fi concept art style.”

4. Offensive Images and Racial Bias

4.1 Why Can AI Produce Offensive or Racist Content?

AI depends on the images and text it was trained on. If those training materials have historical or ongoing biases (e.g., stereotypes about certain groups), the AI can reproduce them.

Example of a Racist Stereotype: An AI might create an image that shows “Black people eating watermelons” if it learned from old cartoons or racist internet sources that perpetuated that stereotype (a racist stereotype rooted in post-Civil War America, where Black people were depicted as simple-minded and excessively fond of watermelon to mock their freedom and reinforce white supremacist caricatures.)
This is hurtful because it repeats harmful narratives that were used to demean Black communities historically.

4.2 Bias in Professional Roles (CEOs, Doctors, etc.)

If you generate an image using the prompt of “CEO,” many AI models might show a white man in a suit. Why?

The data (images from news articles, business websites, etc.) often depict white men as leaders.
The AI is just reflecting the most common images it has seen.

Activity (Optional with CreatiCode):

Prompt 1: “CEO in a boardroom.” Observe who appears.
Prompt 2: “CEO in a boardroom, female, diverse.” Compare results.
Question to think about:
- “Did you notice a difference in how the CEO is portrayed?”
- “How might these biases affect people seeing these images all the time?”

Takeaway: AI can unintentionally reinforce harmful stereotypes. We need to be aware of these biases, question them, and push for more inclusive prompts and data sets.

5. Misuse and Misinformation

5.1 Deepfakes and Fake News

AI can create realistic images of people doing things they never did or saying things they never said.
This can be used maliciously to frame people for crimes, spread political misinformation, or create fake evidence.

5.2 Why This Matters to You

Even if you’re just having fun, be careful about sharing AI-generated images that look “too real.” They could be misunderstood or used by others to mislead people.

Questions to think about:

“Have you seen any viral images or videos that turned out to be fake?”
“What steps can you take to verify if an image is real or AI-generated?”

6. Putting It All Into Practice: Interactive Exercise

Try the five-step challenge below:

Generate a Neutral Image Prompt
- E.g., “A beautiful sunset over the ocean with clouds shaped like animals.”
- Observe: This prompt is unlikely to cause copyright or offensive issues.
Generate an Image with a Famous Character
- E.g., “Pikachu on a surfboard in a Hollywood action scene.”
- Think: Are you using trademarked content (Pikachu)? What if you share this image widely?
Generate an Inclusive Prompt
- E.g., “A diverse group of CEOs in a boardroom, representing different genders and races.”
- Compare with a simpler prompt: “A CEO in a boardroom.”
- Notice: Differences in representation.
Generate a Possibly Sensitive Prompt
- E.g., “Black people eating watermelons.” (This is an offensive prompt, historically.)
- Warning: You may choose not to actually generate it if you’re uncomfortable — the point is to discuss how harmful stereotypes might appear.
- Reflect: Why is this harmful? How should AI tools handle such requests?
Generate a Fake News Prompt
- E.g., “A well-known political figure stealing money from a bank.”
- Ask: How realistic does the image look? Could someone believe it’s real?

After each prompt, answer these questions:

“Was there any sign of copyright violation or trademark issue?”
“Did I notice any bias?”
“Could this image be misleading if shared online?”
“What ethical considerations come to mind?”

7. Reflection and Next Steps

7.1 Reflection Questions

Personal Take:
- What was the most surprising thing you learned about AI-generated images?
- Did you notice any stereotypes or biases in the images you generated or discussed?
Responsibility and Respect:
- Who should be held responsible if an AI-generated image is offensive— the tool, the user, or the company that made the AI?
- Is it okay to pass off AI-generated images as your own artwork without mentioning AI?
Positive Uses:
- Can you think of a scenario where AI image generation helps people? For example, helping with concept art for a school project or a new invention design?

8. Additional Considerations

Fair Use and Transformative Work: In some places, using existing characters or art in a new, creative way might be protected under “fair use.” But fair use is complicated and varies by country.
Privacy Concerns: AI can generate images of real people. How would you feel if your image was used without permission?
Future of AI Regulation: Governments and organizations are starting to create guidelines to manage AI’s impact. Stay updated on new laws or rules that might affect what you can and can’t do with AI-generated art.

Final Thoughts

AI image generation is an exciting and fast-growing field. It offers amazing possibilities for creativity but also poses serious ethical questions. By understanding how AI “remixes” existing art, being aware of potential copyright issues, recognizing biases, and watching out for misuse or misinformation, you can become a more responsible user of this technology.

Remember: As AI becomes a bigger part of everyday life, your awareness and thoughtful approach will help shape a future where AI is used fairly, respectfully, and ethically.