Ethics, Privacy, Copyright and AI Tools Ecosystem

Gains:

• Be able to frame the use of artificial intelligence responsibly in terms of design originality, copyright, data privacy and dark pattern ethics
• Ability to select Figma, Miro and large language model-based UX tools according to the task and integrate them into the workflow
• Ability to create a personal and corporate artificial intelligence usage policy and verification checklist

Throughout this module, we used AI as an accelerator at every stage of the UX workflow. This final unit puts all this power into a responsible framework: ethical boundaries, data privacy, copyright and intellectual property, and choosing the right tool for the right job. For a designer, these are not "extra attention" but the core of the profession; because design directs the behavior of millions of people and this power can be abused. AI does not eliminate this responsibility, it increases it: when a tool performs rapid production on your behalf, the ethical and legal outcome of the output is still yours.

Dark patterns: abuse of power

A dark pattern is design that tricks the user into taking an action they don't want: hiding the cancel button, writing "no" in an embarrassing way ("No, I don't want to save"), deliberately making it difficult to unsubscribe, pre-ticked checkboxes. These inflate metrics in the short term, but destroy user trust and are increasingly subject to legal sanction (consumer protection regulations prohibit them in many countries).

AI can make it easier for you to produce a dark pattern; A request like "make the cancel button unnoticeable" can technically be answered. But this request is unethical and should be rejected. The vehicle is not an excuse; The responsibility always lies with the designer. The litmus test for ethical design is simple: “If the user understood this design, would they still accept it?”

Caution: "AI suggested" is not a defence. The responsibility for an unethical output lies with the designer who uses it in the product.

Data privacy and KVKK: from start to finish

The principle we have been emphasizing since the first unit comes into full picture here. UX research works with personal data and KVKK protects this data. Three rules remain constant when using artificial intelligence:

1. Anonymize: Mask identifiers such as name, contact, workplace.
2. Stay within the limits of consent: Do not exceed the intended use you told the participant.
3. Know the tool's data policy: Control where data is stored and whether it goes to model training; Choose corporate approved tools.

Entering critical data into a free tool could expose the data to third parties. In the corporate environment, be sure to use tools approved by the information security team.

Copyright, intellectual property and originality

Images, icons and texts produced by artificial intelligence are uncertain in terms of copyright. An image generator may produce output that is overly similar to the artifacts it was trained on; This creates the risk of violating someone else's intellectual property. Three controls are essential for commercial use:

• License terms: Do you have the right to use the output of the tool commercially?
• Similarity risk: Does the output look dangerously similar to an existing brand or work?
• Originality and transparency: Declare artificial intelligence contribution in creative works when necessary; Do not present someone else's work as your own original work.

Originality in education and creative fields is also a professional ethic. AI may be a starting point, but the final work must bear your input, judgment and responsibility.

Tool ecosystem: the right tool for the right job

Artificial intelligence tools are roughly divided into three groups. Neither one is suitable for every job; select by task.

vehicle type

Example usage

His strength is

Attention

Large language model (text)

Research synthesis, UX writing, documentation

Text production and analysis

Hallucination, data privacy

Plugin embedded in design tool (e.g. in-Figma)

Wireframe, content filling, layout

Speed in design flow

Consistency, component bond

Collaboration board plugin (e.g. within Miro)

Affinity diagram, brainstorming

Research synthesis, grouping

human verification

Image/icon generator

Illustration, concept visual

Quick visual idea

Copyright, license, similarity

Practical approach: language model for text and analysis, in-tool plugins for design generation, whiteboard plugins for synthesis, generators for visuals — but in each the validation reflexes of this module apply.

three mini cases

Case 1 — Rejected dark pattern request. A product manager asked the designer to “make the cancellation flow so difficult that no one can cancel.” The designer explained that this is a dark pattern and carries trust and legal risk; instead, he designed an honest flow that offers an alternative to those who cancel. The cancellation rate remained manageable, trust was maintained.

Case 2 — Copyright risk caught. A team wanted to use a logo-like icon produced by artificial intelligence. During the compliance check, the icon was found to be dangerously similar to a well-known brand and was not used. Lesson: visual output is always checked for similarity.

Case 3 — Confidentiality breach prevented. A designer was about to upload client conversations along with personal data to an unapproved tool. Institutional policy prevented this; The data was first anonymised, and a certified tool was used. Lesson: policy and anonymization are applied from the beginning, not afterwards.

Copiable prompts

Evaluate this design request ethically: "<<request>>". Does it mislead the user, restrict freedom of choice, is it a dark pattern? Litmus test: if the user understood this, would they still accept it? If problematic, suggest an honest alternative.

Check this AI image/icon output for commercial use:1) is there a risk of similarity to a well-known brand or work?2) what should I look for in terms of licensing/commercial use?List the points to check and the "obtain legal approval" warning if necessary.Image description: <<text>>

Write us an AI usage policy draft (for the UX team): what data can/cannot be uploaded, anonymization rule, approved tools, output verification step, copyright control, ethical boundaries. Give it in short, actionable bullet points.

Which type of AI tool is best suited for this task? "<<task>>".Options: language model / design tool plugin / whiteboard plugin /image generator. Choose the most appropriate one, justify it and write down the risk to be considered.

Weak prompt / Strong prompt

Weak: “Can I use this image?”

The result: a vague, superficial response that fails to assess risk.

Güçlü: "Check this image output for commercial use: is there a risk of similarity, what should I pay attention to in terms of licensing, is legal approval required?"

Result: Concrete risk checklist and required approval notice.

Difference: strong prompt queries the similarity + license + approval dimensions separately.

Common mistakes

• Avoiding responsibility by saying "Artificial intelligence suggested it". The ethical and legal outcome lies with the designer.
• Mistaking dark patterns for "optimization". Any design that deceives the user destroys trust and reputation.
• Using visual output without copyright control. Similarity and licensing risk can lead to business disaster.
• Entering critical data into an unauthorized vehicle. KVKK violation and data leak.
• Imposing a single tool for every job. Not choosing a vehicle according to the task type reduces efficiency and quality.

In summary

The power of AI does not eliminate responsibility; enlarges. Ethical design refuses to deceive the user; dark patterns sacrifice trust and legal security for short-term gain. Data privacy is protected by anonymization, consent and approved tools within the framework of KVKK. Copyright and originality require similarity and licensing control, especially in visual and creative output. Choose the tool ecosystem according to the task: language model to text, tool plugin to design, board to synthesis, generator to visual — but in each, the verification reflex remains constant. Write your own and your team's AI policy; thus speed is balanced with responsibility.

Application task

1. With the third prompt, produce a brief draft AI usage policy for your UX team.
2. Compare the policy with this module's authentication and privacy policies and fill in any gaps.
3. Put a questionable design request to the ethical litmus test with the first prompt.
4. Check the copyright of any artificial intelligence image you are using or will use with the second prompt.
5. Create a self-verification checklist and hang it on your desk.

checklist

• [ ] I rejected the dark pattern requests and produced an honest alternative.
• [ ] I applied the ethical litmus test (“Would the user accept it if they understood it?”).
• [ ] I anonymized personal data and used only approved tools.
• [ ] I checked the visual outputs for copyright and similarity.
• [ ] I matched the task to the correct vehicle type.
• [ ] I created an AI usage policy and verification list.

Module Exam

1. A UX designer puts the 'users want speed' theme that AI derived from 20 interviews into his research report without looking at any quotes; In the presentation, the stakeholder asks 'in which meeting?' When asked, no basis can be found. What is the main lesson of this situation?

• A) Connecting the artificial intelligence output to the real interview quote and using it without verifying it; ✔ Ignoring responsibility for establishing the evidence trail
• B) It is strictly forbidden to use artificial intelligence in research synthesis
• C) The number of meetings should be 50 instead of 20
• D) The theme is not presented in a table

Description: Artificial intelligence can produce fluent but unfounded themes (hallucinations). Not every synthesis output should go into a report or decision without being verified and linked back to the actual user quote and source; The responsibility lies with the designer.

2. Which of the following tasks uses AI with the highest risk and necessarily requires final approval from the designer?

• A) Producing three different layout variations for a wireframe
• B) Suggest five alternatives for the button text
• C) Making the final decision that an interface is accessible and meets user needs ✔
• D) Summarizing a meeting note

Description: The final decision on accessibility approval and actual user need is high risk as it directly affects people's experience. Producing a draft wireframe or text variation is low risk; The final judgment and approval belongs to the designer.

3. What is the most critical step before uploading the user call transcript to AI?

• A) Upload the transcript as is, with all personal data
• B) Anonymizing personal data (name, contact, workplace) and working within the limits of consent and corporate policy ✔
• C) Translate the transcript into English first
• D) Converting the transcript into a visual

Description: Interview transcripts contain personal data such as name, telephone, and workplace. In accordance with KVKK and consent limits, this data must be anonymized or masked before being transferred to artificial intelligence; The vehicle approved by the institution must be used.

4. A persona produced by artificial intelligence contains a sentence such as 'older users are afraid of technology'. What is the right approach?

• A) Leaving the sentence as is, because AI is neutral
• B) Delete the persona completely
• C) Mark the stereotype, test it with real data and remove it if it is unsupported ✔
• D) Emphasizing the sentence further

Explanation: Artificial intelligence may carry stereotypes and bias from the data it is trained on. Persona must be based on real segment data; Stereotypes should be marked and tested against data, and unsupported generalizations should be eliminated.

5. What does 'affinity mapping' mean in qualitative research synthesis?

• A) Method of reaching themes by grouping similar observations and quotes ✔
• B) A survey measuring users' closeness to each other
• C) Technique for choosing the color palette
• D) A type of prototype animation

Explanation: Proximity diagram is the grouping of individual observations and quotations according to their similarities and turning them into themes. AI can speed up initial grouping, but the meaning of the groups must be human-verified.

6. A designer tells the AI ​​to 'write button text' and gets a one-word result without context. What should be added for a stronger prompt?

• A) Just saying 'write better'
• B) Write the prompt in English
• C) Repeating the same prompt many times
• D) Purpose of the display, user context, brand tone, length limit and to give desired number of variations ✔

Description: The powerful prompt gives context: the purpose of the screen, what the user is doing at the moment, brand tone, length limit and number of variations. Contextless request produces generic and unusable output.

7. How is 'severity' determined when prioritizing usability test findings?

• A) According to the seniority of the person who found the problem
• B) According to the impact of the problem on the user, frequency and business impact ✔
• C) According to the order of the findings in the report
• D) According to the random score given by artificial intelligence

Description: Severity is determined by criteria such as how much the problem affects the user (does it prevent the task from being completed), how many users it has seen, and business impact. AI may suggest draft rankings, but the final weight decision is up to the designer and his team.

8. Which is true about AI accessibility recommendations (e.g. alt text)?

• A) Certification of artificial intelligence guarantees accessibility, no additional testing is required
• B) Accessibility is only about color choice
• C) Alt text is not required
• D) Artificial intelligence produces drafts; true accessibility confirmed by assistive technology and user testing ✔

Description: AI can produce quick sketches for alt text and contrast, but true accessibility is only confirmed by testing with screen readers and actual assistive technology users. Automatic suggestion is preliminary, not proof.

9. In an interface text, the AI ​​suggests an error message 'Your operation failed, try again'. What is the best improvement in terms of UX writing?

• A) Add the technical error code to the message and leave the user alone
• B) Remove the message completely
• C) Write in clear and non-accusatory language what happened, the possible reason and the concrete step the user will take ✔
• D) Writing 'ERROR' in capital letters

Description: A good error message clearly states what happened, why, and what the user should do; It wouldn't be accusatory. The 'what happened + why + next step' formula keeps the user on track. The AI ​​sketch is improved by this criterion.

10. What is the most correct attitude about the wireframe produced by an artificial intelligence tool that generates an interface from text?

• A) Using the output directly as the final design
• B) Not using the output at all because it is worthless
• C) Using the output for its colors only
• D) Treat the output as a quick start draft and mature it with content, edge case and accessibility ✔

Description: These tools produce a quick start draft, but are lacking in things like actual content, edge cases, accessibility, and mental models. The output is a starting point; The designer should criticize and mature this.

11. What is the most appropriate use of artificial intelligence in design system work?

• A) Produce component documentation and usage text drafts and verify them by the team ✔
• B) Automatically publish the entire design system without approval
• C) Changing brand colors without asking the designer
• D) Deleting components from the system without user testing

Description: AI is powerful at quickly generating component documentation, usage scripts, do/don't examples, and token naming drafts. However, their proposals should be checked for conflict with the current system; The decision of singularity and consistency lies with the team.

12. A designer asks the AI ​​to 'make the cancel button unnoticeable' for a hidden design that redirects the user to an unwanted subscription. What kind of problem is this?

• A) A harmless micro-interaction improvement
• B) Accessibility improvement
• C) A standard design system rule
• D) A dark pattern that misleads the user, is unethical and carries legal risk ✔

Description: Designs that mislead the user into an undesirable action are 'dark patterns', which are unethical and increasingly subject to legal sanctions. Although AI is an agent, the responsibility lies with the designer; such request should be rejected.

13. What control is required before using an image or icon set produced with artificial intelligence in the product?

• A) No controls required, AI output is free
• B) Verification of the vehicle's license terms, copyright and risk of proximity to similar works ✔
• C) Checking file size only
• D) Counting the number of colors of the image

Description: Artificial intelligence images may be uncertain in terms of license, copyright and brand similarity. In commercial use, the license terms of the tool, excessive proximity to similar works, and intellectual property risk should be verified; If necessary, legal/compliance approval must be obtained.

14. What is the right approach about the 'emotion' line produced by AI on the user journey map?

• A) Confirming emotions with real research data, marking unfounded ones as assumptions ✔
• B) Considering artificial intelligence's emotion prediction as absolute truth
• C) Completely ignoring the emotion line
• D) Marking emotions with random colors

Explanation: AI can reasonably predict emotions and pain points, but these are assumptions unless supported by real user data. The map should be confirmed by research citations; unfounded feelings should be marked as assumptions.