AI Ethics Risk Intelligence Dashboard: EU AI Act & Ethical Risk Analysis of AI Incidents
GOAL
RESULT
PROJECT DURATION
The objective of this project was to design an interactive Tableau dashboard that analyzes real-world AI incidents through ethical risk dimensions, harm categories, and EU AI Act-oriented risk classifications.
The project focuses on the intersection of data analytics, AI ethics, and AI governance. Instead of looking at AI incidents only as isolated events, this dashboard organizes them into a structured analytical framework. The aim is to understand where AI-related harms appear, which sectors are more exposed to ethical risk, and how incidents can be mapped to AI Act-oriented risk categories.
The main analytical question behind the project is:
“How can publicly reported AI incidents be analyzed through ethical risk dimensions and EU AI Act-oriented risk categories?”
This project combines three analytical areas:
Data Analytics and Business Intelligence
Incident-level analysis, sector grouping, harm category analysis, heatmaps, KPI cards, filters, and interactive Tableau dashboards.
AI Ethics
Privacy risk, bias risk, transparency risk, accountability risk, safety risk, autonomy risk, and ethical risk scoring.
AI Governance / EU AI Act-Oriented Analysis
Risk category mapping, prohibited / high-risk incident identification, compliance priority, and governance-focused interpretation.
The final result is a three-dashboard Tableau project that provides a complete view of AI incident risk from executive, exploratory, and governance perspectives.
The dashboard includes:
- Executive AI ethics risk overview
- Total incident and critical incident KPIs
- Average ethical risk score out of 30
- AI incidents by year
- Incidents by sector group
- EU AI Act-oriented risk category distribution
- Ethical risk heatmap by sector and risk dimension
- Incident-level explorer with source information
- Top 10 highest-risk AI incidents
- Ethical risk dimension breakdown
- AI Act risk category vs harm category group heatmap
- AI Act-oriented risk categories by sector group
- Interactive filters for year, sector, AI Act risk, harm category, and priority
The dashboard makes AI ethics and AI governance concepts easier to explore through a structured data analytics workflow.
The final Tableau workbook contains three main dashboards:
- AI Ethics Risk Intelligence Dashboard
- AI Incident Explorer
- EU AI Act & Ethics Analysis
The project was completed within an estimated time frame of 7 to 10 days.
This included:
- Designing the project idea
- Collecting and reviewing public AI incident examples
- Preparing a curated dataset of 200 AI incidents
- Creating source and reference fields
- Cleaning and structuring the Excel dataset
- Creating ethical risk dimensions
- Coding EU AI Act-oriented risk categories
- Grouping sectors and harm categories
- Building Tableau calculated fields
- Designing three interactive dashboards
- Testing filters and dashboard interactions
- Preparing the project for website and portfolio presentation
The dashboard was developed using Tableau Public and Excel.
CASE – AI ETHICS RISK INTELLIGENCE DASHBOARD
In this project, I created a Tableau dashboard to analyze public AI incidents through ethical risk scoring and EU AI Act-oriented risk mapping.
The scenario is based on a curated dataset of publicly reported AI incidents. Each incident represents a case where an AI system caused, contributed to, or was associated with a potential harm, ethical concern, or governance issue.
The project was designed to go beyond a simple incident list. Instead, each incident was analyzed and classified according to:
- Sector
- Harm category
- AI Act-oriented risk category
- Compliance priority
- Ethical risk dimensions
- Overall ethical risk score
- Source URL
The aim was not only to visualize where AI incidents happen, but also to understand how different types of harm relate to broader ethical and regulatory risks.
A traditional incident dashboard might only count incidents by year or category. This project goes further by asking:
- Which sectors show higher ethical risk exposure?
- Which harm categories are most common?
- Which incidents are mapped as high-risk or prohibited?
- Which ethical dimensions appear most strongly across sectors?
- How do AI Act-oriented categories relate to harm types?
- Which incidents should receive higher governance attention?
Business / Research Problem
AI systems are increasingly used in law enforcement, healthcare, finance, marketing, education, public administration, social media, and consumer technology. While these systems can improve efficiency and decision-making, they can also create serious risks.
Examples include:
- Facial recognition errors
- Algorithmic discrimination
- Privacy and surveillance violations
- Deepfake and synthetic media abuse
- AI-generated misinformation
- Healthcare and safety harms
- Fraud or financial model risk
- Lack of transparency and accountability
The main problem is that AI incidents are often discussed separately, without a structured analytical framework.
This project was designed to answer:
How can AI incidents be organized, scored, and visualized in a way that supports ethical and governance-focused analysis?
The project connects my academic background in AI ethics and EU AI governance with practical Tableau dashboarding and data analysis.
Dataset Collection
The dataset used in this project is a curated sample of 200 publicly reported AI incidents.
The data was collected and structured from public AI incident references, source links, news reports, and AI governance-related incident documentation. The aim was not to create a complete global census of all AI incidents, but to prepare a credible portfolio dataset that could support ethical risk analysis.
During the data collection process, I used:
- Public AI incident databases
- Public incident source URLs
- AI governance reports and case references
- News articles and external documentation
- Manual review of incident descriptions
- Excel for data structuring and cleaning
- Tableau Public for dashboard development
The dataset includes incidents related to:
- Facial recognition
- Predictive policing
- Healthcare AI
- AI chatbots
- Generative AI
- Synthetic media
- Deepfakes
- Fraud detection
- Financial risk models
- Marketing and consumer AI
- Public administration systems
- Social media and misinformation
- Autonomous systems and robotics
Each row in the dataset represents one incident.
Dataset Fields
The dataset includes the following fields:
- Incident Id
- Incident Year
- Incident Title
- Sector
- Sector Group
- Harm Category
- Harm Category Group
- Risk Category
- Compliance Priority
- Accountability Risk
- Transparency Risk
- Autonomy Risk
- Privacy Risk
- Safety Risk
- Bias Risk
- Ethical Risk Score
- Risk Score Label
- Source URL
- Analyst Notes
- Documentation Required
- Human Oversight Required
- Transparency Required
The Source URL field was preserved so that incident-level records could be traced back to public references.
Data Collection and Source Review
The data collection process followed a manual review workflow.
First, I identified publicly reported AI incident examples from public AI incident resources, source links, and external reports. Then, I reviewed the incident descriptions and extracted the main analytical fields needed for the dashboard.
For each incident, I captured:
- What happened
- Which sector was involved
- What type of AI system was involved
- What type of harm occurred
- Which group was affected
- What ethical principle was most relevant
- Whether the incident appeared to involve high governance priority
- Which public source documented the incident
After collecting the incident data, I manually enriched the dataset with ethical and governance fields.
This included:
- Assigning harm categories
- Creating broader harm category groups
- Assigning AI Act-oriented risk categories
- Scoring six ethical risk dimensions
- Creating an overall ethical risk score
- Assigning compliance priority levels
This approach allowed the dashboard to move beyond descriptive reporting and support structured ethical risk analysis.
Data Preparation in Excel
After collecting the incident data, I prepared the dataset in Excel.
The preparation process included:
- Cleaning incident titles
- Standardizing year values
- Checking missing values
- Reviewing duplicate incidents
- Standardizing source URLs
- Creating consistent sector labels
- Creating broader sector groups
- Creating broader harm groups
- Checking risk score values
- Preparing fields for Tableau filters
A key part of the data preparation process was grouping overly detailed categories into cleaner analytical categories.
For example, detailed sector values were grouped into broader sector groups such as:
- Law Enforcement / Justice
- Healthcare
- Finance / Insurance
- Media / Social Platforms
- Marketing / Consumer
- Public Administration
- Biometrics / Surveillance
- Autonomous Systems / Transport
- Cybersecurity
- Education
- Employment
- Generative AI / Chatbots
- Other
This made the Tableau dashboard more readable and prevented the charts from becoming too fragmented.
Ethical Risk Scoring
To make the incident analysis more systematic, each incident was evaluated across six ethical risk dimensions:
- Accountability Risk
- Transparency Risk
- Autonomy Risk
- Privacy Risk
- Safety Risk
- Bias Risk
The selection of these dimensions was informed by widely recognized AI ethics and governance frameworks, including:
- European Commission High-Level Expert Group on AI, Ethics Guidelines for Trustworthy AI (2019)
- OECD AI Principles (2019)
- NIST AI Risk Management Framework (AI RMF 1.0, 2023)
These frameworks consistently emphasize principles such as accountability, transparency, human agency and oversight, privacy and data governance, technical robustness and safety, and fairness/non-discrimination.
Each dimension was scored on a scale from 1 (low concern) to 5 (high concern) based on the characteristics of the incident and the severity of the reported ethical issues.
The overall ethical risk score was calculated as the sum of the six dimensions:
Ethical Risk Score =
Accountability Risk
+ Transparency Risk
+ Autonomy Risk
+ Privacy Risk
+ Safety Risk
+ Bias RiskBecause there are six dimensions and each dimension has a maximum value of 5, the maximum total score is:
6 × 5 = 30Therefore, the total ethical risk score ranges from 6 to 30.
For interpretation purposes, the following bands were used:
Low total ethical risk: 6–12
Medium ethical risk: 13–20
High ethical risk: 21–26
Critical ethical risk: 27–30These thresholds were created as an analytical categorization scheme for this portfolio project and are not derived from an official regulatory methodology.
This scoring framework should therefore be understood as an analyst-coded assessment model inspired by established AI ethics and governance principles rather than a formal legal, regulatory, or compliance classification system.
References
European Commission High-Level Expert Group on Artificial Intelligence. (2019). Ethics Guidelines for Trustworthy AI.
OECD. (2019). OECD Principles on Artificial Intelligence.
National Institute of Standards and Technology (NIST). (2023). AI Risk Management Framework (AI RMF 1.0)
Tableau Calculated Fields
Several calculated fields were created in Tableau to make the dashboard cleaner, more readable, and more analytical.
Sector Group
This calculated field groups detailed sector names into broader sector categories.
IF ISNULL([Sector])
THEN
"Other" ELSEIF CONTAINS(LOWER([Sector]),
"employment") OR CONTAINS(LOWER([Sector]),
"hiring") OR CONTAINS(LOWER([Sector]),
"workplace") THEN "Employment" ELSEIF CONTAINS(LOWER([Sector]),
"education") OR CONTAINS(LOWER([Sector]),
"school") OR CONTAINS(LOWER([Sector]),
"student") THEN "Education" ELSEIF CONTAINS(LOWER([Sector]),
"health") OR CONTAINS(LOWER([Sector]),
"medical") OR CONTAINS(LOWER([Sector]),
"clinical") OR CONTAINS(LOWER([Sector]),
"wellbeing") THEN "Healthcare" ELSEIF CONTAINS(LOWER([Sector]),
"finance") OR CONTAINS(LOWER([Sector]),
"bank") OR CONTAINS(LOWER([Sector]),
"credit") OR CONTAINS(LOWER([Sector]),
"insurance") THEN "Finance / Insurance" ELSEIF CONTAINS(LOWER([Sector]),
"law") OR CONTAINS(LOWER([Sector]),
"police") OR CONTAINS(LOWER([Sector]),
"justice") OR CONTAINS(LOWER([Sector]),
"court") OR CONTAINS(LOWER([Sector]),
"immigration") THEN "Law Enforcement / Justice" ELSEIF CONTAINS(LOWER([Sector]),
"biometric") OR CONTAINS(LOWER([Sector]),
"facial") OR CONTAINS(LOWER([Sector]),
"surveillance") THEN "Biometrics / Surveillance" ELSEIF CONTAINS(LOWER([Sector]),
"marketing") OR CONTAINS(LOWER([Sector]),
"advertising") OR CONTAINS(LOWER([Sector]),
"consumer") THEN "Marketing / Consumer" ELSEIF CONTAINS(LOWER([Sector]),
"media") OR CONTAINS(LOWER([Sector]),
"social") OR CONTAINS(LOWER([Sector]),
"content") OR CONTAINS(LOWER([Sector]),
"publishing") THEN "Media / Social Platforms" ELSEIF CONTAINS(LOWER([Sector]),
"transport") OR CONTAINS(LOWER([Sector]),
"vehicle") OR CONTAINS(LOWER([Sector]),
"autonomous") OR CONTAINS(LOWER([Sector]),
"robot") THEN "Autonomous Systems / Transport" ELSEIF CONTAINS(LOWER([Sector]),
"public") OR CONTAINS(LOWER([Sector]),
"government") OR CONTAINS(LOWER([Sector]),
"welfare") OR CONTAINS(LOWER([Sector]),
"administration") THEN "Public Administration" ELSEIF CONTAINS(LOWER([Sector]),
"cyber") OR CONTAINS(LOWER([Sector]),
"security") OR CONTAINS(LOWER([Sector]),
"software") THEN "Cybersecurity" ELSEIF CONTAINS(LOWER([Sector]),
"generative") OR CONTAINS(LOWER([Sector]),
"chatbot") OR CONTAINS(LOWER([Sector]),
"llm")
THEN "Generative AI / Chatbots"
ELSE "Other"
ENDWhy this field was used
The original sector values were too detailed and inconsistent for dashboard-level analysis. This calculated field allowed the incidents to be grouped into cleaner sector categories.
It helped make charts such as Incidents by Sector Group and Average Ethical Risk by Sector and Risk Dimension easier to read.
Harm Category Group
This calculated field groups detailed harm categories into broader ethical harm groups.
IF ISNULL([Harm Category])
THEN "Other"
ELSEIF CONTAINS(LOWER([Harm Category]),"bias")
OR CONTAINS(LOWER([Harm Category]), "discrimination")
THEN "Bias / Discrimination"
ELSEIF CONTAINS(LOWER([Harm Category]), "privacy")
OR CONTAINS(LOWER([Harm Category]), "surveillance")
OR CONTAINS(LOWER([Harm Category]), "civil liberties")
OR CONTAINS(LOWER([Harm Category]), "rights") THEN "Privacy / Surveillance / Rights"
ELSEIF CONTAINS(LOWER([Harm Category]), "fraud")
OR CONTAINS(LOWER([Harm Category]), "financial")
OR CONTAINS(LOWER([Harm Category]), "model risk") THEN "Fraud / Financial Harm"
ELSEIF CONTAINS(LOWER([Harm Category]), "misinformation")
OR CONTAINS(LOWER([Harm Category]), "influence")
OR CONTAINS(LOWER([Harm Category]), "manipulation")
THEN "Misinformation / Manipulation"
ELSEIF CONTAINS(LOWER([Harm Category]), "physical")
OR CONTAINS(LOWER([Harm Category]), "safety")
OR CONTAINS(LOWER([Harm Category]), "health")
OR CONTAINS(LOWER([Harm Category]), "self-harm")
THEN "Safety / Health Harm"
ELSEIF CONTAINS(LOWER([Harm Category]), "cyber")
OR CONTAINS(LOWER([Harm Category]), "operational")
THEN "Cybersecurity / Operational Harm"
ELSEIF CONTAINS(LOWER([Harm Category]), "sexual")
OR CONTAINS(LOWER([Harm Category]), "intimate")
THEN "Synthetic Sexual Abuse"
ELSEIF CONTAINS(LOWER([Harm Category]), "education")
OR CONTAINS(LOWER([Harm Category]), "employment")
THEN "Education / Employment Harm"
ELSE "Other AI-related Harm"
ENDWhy this field was used
The original harm categories were too detailed for high-level analysis. This field grouped them into broader ethical harm patterns.
It was used in:
- Incidents by Harm Category Group
- AI Act Risk Category vs Harm Category Group heatmap
- Incident Explorer filters
- Governance analysis dashboard
Risk Category Order
This field was created to sort AI Act-oriented risk categories by severity instead of alphabetical order.
CASE [Risk Category]
WHEN "Prohibited" THEN 1
WHEN "High / Product Safety" THEN 2
WHEN "High" THEN 3
WHEN "High / Unclear" THEN 4
WHEN "Needs Review" THEN 5
WHEN "Limited / Unclear" THEN 6
WHEN "Limited" THEN 7
WHEN "Minimal / Business Risk" THEN 8
ELSE 9
ENDWhy this field was used
Tableau normally sorts text categories alphabetically. That would make the risk categories difficult to interpret.
This field allowed the categories to be sorted from the most serious to the least serious:
Prohibited
High / Product Safety
High
High / Unclear
Needs Review
Limited / Unclear
Limited
Minimal / Business RiskThis sorting was used in AI Act-oriented visuals and heatmaps.
Risk Score Text
This field converts the ethical risk score into a clean text format for tables.
STR(ROUND([Ethical Risk Score], 0))STR(ROUND([Ethical Risk Score], 0))[Ethical Risk Score], 0))Why this field was used
In Tableau text tables, numeric measures sometimes appear as axes or aggregated marks. This field allowed the risk score to appear as a clean table value.
In the final dashboard, the column title was shown as:
Risk Score / 30This makes it clear that the score is based on a maximum total risk score of 30.
Short Incident Title
This field shortens long incident titles for better dashboard readability.
IF LEN([Incident Title]) > 70
THEN LEFT([Incident Title], 70) + "..."
ELSE [Incident Title]
ENDWhy this field was used
Many AI incident titles are long and descriptive. Long titles can make tables and dashboards difficult to read.
This field was used in the Top 10 incident list and incident-level tables to keep the dashboard clean.
Total Incidents
This field counts the number of unique AI incidents.
IF LEN([Incident Title]) > 70
THEN LEFT([Incident Title], 70) + "..."
ELSE [Incident Title]
ENDWhy this field was used
This metric was used in KPI cards and summary views to show the total number of incidents in the curated dataset.
Critical Incidents
This calculated field counts incidents with critical compliance priority.
SUM(
IF [Compliance Priority] = "Critical" THEN 1 ELSE 0 END
)Why this field was used
This KPI highlights incidents that require the highest level of governance attention.
Prohibited or High-Risk Incidents
This calculated field counts incidents mapped as prohibited or high risk.
SUM(
IF [Risk Category] = "Prohibited"
OR CONTAINS([Risk Category], "High")
THEN 1
ELSE 0
END
)Why this field was used
This metric was used to identify the number of incidents that fall into the most serious AI Act-oriented risk categories.
Blank
This calculated field removes Tableau’s default “Abc” placeholder in text tables.
""Why this field was used
When building text tables in Tableau, the software can display an unnecessary “Abc” placeholder. This field helped create cleaner incident tables.
Tableau Dashboard Creation
After preparing the dataset and calculated fields, I created three Tableau dashboards.
The dashboards were designed to move from summary analysis to incident-level exploration and then to governance-focused interpretation.
Dashboard 1 – AI Ethics Risk Intelligence Dashboard
The first dashboard provides an executive-level overview of the AI incident dataset.
The first dashboard provides an executive-level overview of the AI incident dataset.
It includes the following KPI cards:
- Total Incidents
- Critical Incidents
- Average Ethical Risk Score / 30
The dashboard also includes the following visuals:
Incidents by Sector Group
This horizontal bar chart shows the number of AI incidents by sector group.
The visual helps identify which sectors are more represented in the curated incident dataset. It also makes the dashboard easier to understand by using grouped sectors instead of detailed raw sector names.
AI Incidents by EU AI Act-Oriented Risk Category
This visual shows how incidents are distributed across AI Act-oriented risk categories such as:
- Prohibited
- High
- High / Unclear
- High / Product Safety
- Limited
- Limited / Unclear
- Needs Review
- Minimal / Business Risk
The purpose of this chart is to connect incident-level analysis to governance-oriented risk classification.
Incidents in Curated Dataset by Year
This chart shows how incidents are distributed by year in the curated dataset.
The chart should not be interpreted as a complete global trend of all AI incidents. Instead, it shows the year distribution within the curated sample used for this portfolio project.
Average Ethical Risk by Sector and Risk Dimension
This heatmap compares sector groups across six ethical risk dimensions:
- Accountability Risk
- Transparency Risk
- Autonomy Risk
- Privacy Risk
- Safety Risk
- Bias Risk
The purpose of this visual is to show which sectors are associated with higher average ethical risk across different dimensions.
Dimension scores range from 1 to 5.
Dashboard 2 – AI Incident Explorer
The second dashboard allows users to explore individual AI incidents and source-level details.
It includes:
- Incident details and sources
- Top 10 highest-risk AI incidents
- Ethical risk dimension breakdown
- Incidents by harm category
- Interactive filters
The page includes filters for:
- Year
- Sector
- AI Act Risk
- Harm Category
- Compliance Priority
Incident Details and Sources
This table allows users to explore incident-level data.
It includes:
- Year
- Incident
- Sector
- Harm Category
- Risk Category
- Risk Score / 30
The purpose of this table is to create transparency and allow users to move from dashboard-level patterns to individual incident examples.
Top 10 Highest-Risk AI Incidents
This table shows the highest-risk incidents based on the ethical risk score.
The risk score is shown out of 30.
This visual helps users identify the most severe incidents in the curated dataset.
Ethical Risk Dimension Breakdown
This bar chart shows the average scores across the six ethical risk dimensions.
The chart helps users understand which ethical dimensions are most prominent across the incident sample.
The dimension scores are based on a 1 to 5 scale.
Incidents by Primary Harm Category
This chart shows the most common harm categories in the dataset.
It highlights which types of harm appear most frequently, such as:
- Fraud / financial loss
- Privacy / surveillance
- Physical safety
- Bias / discrimination
- Misinformation / influence
- Synthetic or intimate-image abuse
Dashboard 3 – EU AI Act & Ethics Analysis
The third dashboard focuses on AI governance and EU AI Act-oriented analysis.
It includes:
- Average ethical risk score by AI Act-oriented category
- AI Act risk category vs harm category group heatmap
- AI Act-oriented risk categories by sector
Average Ethical Risk Score by AI Act-Oriented Category
This chart compares average ethical risk score across AI Act-oriented categories.
It helps evaluate whether more serious AI Act-oriented categories are associated with higher analyst-coded ethical risk scores.
The score is measured out of 30.
AI Act Risk Category vs Harm Category Group
This heatmap shows how harm categories intersect with AI Act-oriented risk categories.
For example, it helps identify where harms such as fraud, privacy / surveillance, safety / health harm, misinformation, and synthetic sexual abuse appear within different risk categories.
This visual is one of the central governance-focused visuals in the project.
AI Act-Oriented Risk Categories by Sector
This chart shows how AI Act-oriented risk categories are distributed across sectors.
The purpose is to understand which sectors contain more high-risk, prohibited, limited, or unclear cases.
Key Insights
Several important insights emerged from the dashboard.
1. Ethical risk is not concentrated in only one sector
The dashboard shows that AI incidents appear across several sectors, including law enforcement, healthcare, finance, media, marketing, public administration, education, and consumer technology.
This suggests that AI ethics risk is not limited to one industry.
2. Facial recognition and surveillance-related incidents show high governance concern
Incidents involving facial recognition, biometric systems, policing, surveillance, and identification systems often show elevated privacy, bias, transparency, and accountability risks.
These incidents are frequently connected to high-risk or prohibited categories.
3. Synthetic media and generative AI incidents create distinct harm patterns
Deepfake, synthetic media, and AI-generated content incidents are strongly connected to misinformation, manipulation, fraud, and sexual abuse-related harms.
These cases show why generative AI governance needs both technical and ethical monitoring.
4. Healthcare-related incidents show high safety and accountability relevance
Healthcare AI incidents are especially important because errors or misleading outputs may directly affect human wellbeing.
This makes safety, accountability, and oversight particularly important in healthcare-related AI use cases.
5. Transparency and accountability are consistently high ethical risk dimensions
Across many sectors and risk categories, transparency and accountability appear as recurring ethical concerns.
This shows the importance of explainability, documentation, governance review, and human oversight in AI systems.
Limitations
This dashboard uses a curated sample of 200 publicly reported AI incidents.
It should not be interpreted as a complete global database of all AI incidents.
The ethical risk scores, harm categories, compliance priority values, and EU AI Act-oriented mappings are analyst-coded for portfolio and educational purposes.
They do not constitute legal advice or official regulatory classification.
The dashboard is intended to demonstrate how data analytics can support AI ethics and governance analysis
Tools Used
The project was built using the following tools:
- Tableau Public
- Excel
- Data cleaning
- Manual incident review
- Public AI incident sources
- Ethical risk scoring
- EU AI Act-oriented risk mapping
- Dashboard design
- Data visualization
- AI governance analysis
How to Use the Dashboard
Users can explore the dashboard by navigating through the three Tableau dashboards.
1. AI Ethics Risk Intelligence Dashboard
Use this dashboard to understand the overall ethical risk profile of the incident dataset.
It answers:
- How many AI incidents were analyzed?
- How many incidents are critical?
- What is the average ethical risk score?
- Which sectors appear most often?
- How are incidents distributed across AI Act-oriented risk categories?
- Which sectors show higher ethical risk by dimension?
2. AI Incident Explorer
Use this dashboard to explore individual incidents.
It answers:
- Which incidents have the highest risk score?
- What harm categories appear most often?
- Which incidents belong to each sector?
- What is the ethical risk profile across six dimensions?
- How can incidents be filtered by year, sector, risk category, harm category, and priority?
3. EU AI Act & Ethics Analysis
Use this dashboard to analyze the governance perspective.
It answers:
- Which AI Act-oriented categories have higher average risk?
- How do harm categories intersect with risk categories?
- Which sectors contain more high-risk or prohibited incidents?
- How can ethical risk scoring support AI governance analysis?
Conclusion
The AI Ethics Risk Intelligence Dashboard demonstrates how real-world AI incidents can be analyzed through a structured data analytics and AI governance framework.
The project shows that AI ethics can be translated into measurable and visual risk indicators using Tableau.
Instead of presenting AI incidents only as separate examples, the dashboard organizes them into ethical dimensions, harm categories, sector groups, and AI Act-oriented categories.
This project connects my background in data analysis and dashboard development with my academic interest in AI ethics and EU AI governance.
It is especially relevant for roles involving:
- Data Analysis
- Business Intelligence
- AI Governance
- Responsible AI
- Risk Analytics
- Policy Analytics
- Product Analytics
- Technology Ethics
A good AI governance workflow should not only identify where AI systems are used. It should also monitor where harms appear, which groups may be affected, and which types of systems require stronger accountability, transparency, and oversight.
