• Understand basic AI definitions and concepts, including how AI differs from traditional tools; 

• Articulate AI's potential business value through simple examples; 

• Develop curiosity towards AI's capabilities while maintaining realistic expectations; 

• Comprehend basic AI regulatory frameworks and their relevance to business operations; 

• Identify common compliance requirements and apply basic checks (e.g., risk assessments); 

• Prioritize ethical and legal considerations in AI adoption to build trust and avoid penalties;

• Understand key principles of fairness, transparency, and privacy in AI use.

• Identify responsible practices and apply simple safeguards (e.g., reviewing outputs, protecting sensitive data).

• Value accountability and maintain trust in how AI is applied.

Definition:

AI as systems that mimic human intelligence to perform tasks like recognizing patterns or making predictions.

Analogy:

Compare AI to a recipe-following chef that learns from past meals.

Myths vs. Realities:

Myth—"AI will replace all jobs“.
Reality—AI augments human work, freeing time for creativity in MSMEs (Deloitte "Tech Trends" 2023).

Healthcare

Companion robots for the care of the elderly

Mining medical records to provide more useful information

Design treatment plans

Assist in repetitive jobs including medication management

Provide consultations

Using avatars in place of patients for clinical training

Heavy Industry

Robots have become common in many industries and
are often given jobs that are considered dangerous to humans.

Robots have proven effective in jobs that are very repetitive which may lead
to mistakes or accidents due to a lapse in concentration and other jobs
which humans may find degrading.

Finance

Algorithmic

Trading

Market Analysis and Data Mining

Personal Finance

Portfolio Management Underwriting

• More powerful and more useful computers.
• New and improved interfaces.
• Solving new problems.
• Better handling of information.
• Relieves information overload.
• Conversion of information into knowledge.

• Increased costs
• Difficulty with software development slow and expensive
• Few experienced programmers
• Few practical products have reached the market as yet.

AI Model

Input

Output

AI technology is not new. You probably already use it in your everyday life, without thinking about it.

For example, when: 
Spotify suggests music you might like, 

• Google Maps shows you the fastest route, 
• Your phone sorts photos by people. 
   

What's new about generative AI is that it doesn't just analyse - it creates. And it opens up many new ways to get help in the work!

How does AI generate anything new, and what do you need to know to grasp how it operates in the background? Here's an in-depth look at how to use AI intelligently and safely.

As previously stated, ChatGPT and Copilot use language models that have been trained on massive amounts of text, such as books, papers, and web pages. They learn to identify patterns in how people write and communicate. When you ask a question, the AI attempts to guess a possible response—it has no idea what's correct or wrong.

Consider it the world's most advanced autocomplete. A language model, like your mobile phone, proposes the next word as you enter a message - but on a much wider scale and with a better comprehension of the context.

Keep in mind!
AI doesn't have the understanding of a human.
It lacks consciousness, emotions and its own experiences.

AI predicts responses based on what it has seen most of – not what is true.
It reproduces the patterns it has learned are statistically probable.

AI can sometimes be wrong or even make things up.
These are called hallucinations and occur when the model generates information that does not correspond to reality.

AI does not understand causation or ethics.
It can generate texts on these topics, but has no real insight into why something is happening or what is morally right. After basic training, many models get a second phase: humans review the answers and teach the AI which answers are helpful and harmless. This acts as an imposed safety filter, but it is a learned behavior – not an ethical understanding.

It might not always be clear how generative AI may help with your specific task, but a smart first step is to ask yourself:

What takes up a lot of time in my daily life?

Are there any sections that are repetitious, time-consuming, or difficult to get started on?

Then AI can be useful.

Automation:

Fixed, rule-based processes, e.g., a spreadsheet auto-summing sales in a MSME.

Artificial Intelligence (AI):

Adaptive and learning-based, learns from data to handle variability, e.g., chatbots adapting to customer queries.

Why regulate?
Regulations ensure safe AI use; for European MSMEs, compliance avoids fines (up to 4% of revenue under GDPR).
Focus: EU AI Act, GDPR, and practical steps for small businesses.

- 75% of SMEs report regulatory awareness as key to adoption (European Court of Auditors "Special Report" 2024).

• Al regulations varied significantly across different countries and regions, with some jurisdictions taking more proactive steps than others.
• Here are some examples of current Al regulations and initiatives.

European Union (EU)

• The EU has proposed the Artificial Intelligence Act (AI Act), a comprehensive regulatory framework aimed at governing the development and deployment of Al systems within the EU.
• The proposed regulation categorizes Al systems based on their risk levels and imposes requirements for transparency, accountability, and human oversight.
• The General Data Protection Regulation (GDPR), although not specifically focused on Al, establishes data protection principles that apply to Al systems, particularly concerning the processing of personal data.AI Act

United States

• As of now (October 2025), the United States lacks comprehensive federal regulations specifically targeting Al.
• However, various federal agencies, including the Federal Trade Commission (FTC) and the National Institute of Standards and Technology (NIST), have issued guidance documents and reports on Al-related topics such as fairness, transparency, and accountability.
• Some states, such as California, have enacted legislation addressing specific Al-related issues, including privacy (e.g., California Consumer Privacy Act) and automated decision-making (e.g., Automated Decision Systems Accountability Act).

Canada

Canada has established the Pan-Canadian Artificial Intelligence Strategy, a federal initiative aimed at fostering Al research and innovation in Canada.

The strategy includes investments in Al research, talent development, and commercialization efforts, as well as initiatives to address ethical and societal implications of Al.

China

• China has issued guidelines and standards for Al development, focusing on areas such as Al ethics, data security, and industry standards.
• The Chinese government has also implemented regulatory measures to govern Al applications in sectors such as finance, healthcare, and transportation.

International Organizations

International organizations such as the Organisation for Economic Co-operation and Development (OECD) and the United Nations (UN) have developed guidelines and principles for Al ethics and governance.

For example, the OECD has issued the OECD Principles on Al, which provide recommendations for responsible Al development and deployment.

The General Data Protection Regulation (GDPR) is a privacy and security law drafted and passed by the European Union (EU), that imposes obligations onto organizations anywhere.

This goes back to the right to privacy, part of the 1950 European Convention on Human Rights: “Everyone has the right to respect for his private and family life, his home and his correspondence”

• The right to be informed
• The right of access
• The right to rectification
• The right to erasure
• The right to restrict processing
• The right to data portability
• The right to object
• Rights in relation to automated decision making and profiling.

The European Commission appointed a group of experts to provide advice on its artificial intelligence strategy: High-Level Expert Group on AI.

According to the Guidelines, trustworthy AI should be:

• Lawful: respecting all applicable laws and regulations.
• Ethical: respecting ethical principles and values.
• Robust: both from a technical perspective while taking into account its social environment.

• Human agency and oversight: AI systems should empower human beings.
• Technical Robustness and safety: AI systems need to be resilient and secure.
• Privacy and data governance: data governance mechanisms must be ensured.
• Transparency: the data, system and AI business models should be transparent.
• Diversity, non-discrimination and fairness: AI systems should be accessible to all.
• Societal and environmental well-being: AI systems should benefit all human beings.
• Accountability: ensure responsibility and accountability for AI systems and their outcomes.

The AI Act is a document proposed by the European Commission that contains several harmonised rules regarding AI applications;

• emphasising that its approach is shaped by EU values and risk-based, ensuring both safety and fundamental rights protection.

EU AI Act - Regulation COM/2021/206

Goals:

1. To provide a single framework for Al products and services that are placed on the EU market, even where providers are not based in the Union;
2. To ensure that Al systems are safe, secure, trustworthy and respect fundamental rights and values.

• Definition of an Al system: aligned with the OECD; “machine-based system designed to operate with autonomy ... infers, from input received, how to generate output...
• "Scope: applicability for providers, deployers, importers, distributors and manufacturers of Al systems across all EU Member States, with extraterritorial effect similar to the GDPR
• Regulatory approach: modelled on EU product safety legislation, with a risk-based approach
  • Obligations ∞ level of risk

What does the AI Act propose?

• Prohibition of unacceptable AI practices (e.g., social scoring)
• Regulation of high-risk AI systems (e.g., AI used in the context of recruitment)
• Conformity assessment (i.e., under the EU product safety framework)
• Transparency obligations for potentially deceptive AI systems
• Ex post market surveillance (i.e., post-market monitoring system)
• Governance (i.e., authorities must be appointed for the application and implementation)
• Pre-emption of national AI regulatory frameworks (i.e., regulated by the EU)
• Monitoring and enforcement (i.e., done by the Member States)
• Compliance with the prohibitions and regulatory requirements

Unacceptable risk - Al systems that pose a clear threat to the health, safety and rights of people will be prohibited.

• Examples: Al systems deploying subliminal, deceptive techniques with the objective or effect of impairing a person's ability to make an informed decision; biometric categorization systems; social scoring by governments; real-time biometric identification in public spaces.

High risk - Al systems considered to pose a high risk to the health, safety, and rights of people will be subject to strict obligations.

• Examples: Al-powered critical infrastructure; the use of Al in the safety components of products, or the provision of essential public and private services; employment; education; administration of justice. Full list in Annex III.

EU AI Act - Obligations for High Risk Al Systems

• Conformity Assessment
  • Risk identification and mitigation exercise; transparency and accountability; human oversight; data quality.
• Fundamental Rights Impact Assessment
  • For deployers of high-risk Al that are bodies governed by public law or private operators providing public services.
  • Intended purpose of the Al system; duration and frequency of deployment; individuals or groups likely to be affected; specific risks of harm; measures to be taken to mitigate such harms (including governance and complaint mechanisms).

Note: Symbiosis between the different actors identified within the Al Act.

MSME Benefit: Streamlines for small teams, avoiding overlaps.

Practical Example: A MSME creates a matrix linking AI Act and GDPR for e-commerce tools, simplifying audits and enabling secure personalization.

Useful Matrix Example: Columns: Regulation, Requirements, MSME Actions (e.g., GDPR: Consent; Action: Forms).

• The development of data-driven artificial intelligence applications is pushing the limits of the applications of these algorithms in our lives: this rapid evolution motivated the need to ethical, legal and technical regulatory frameworks based on specific principles: accountability, interpretability, fairness, safety, privacy.
• At the European level, there have been several proposals to regulate data and AI-based applications: EU-GDPR, Trustworthy AI Initiative and AI Act.
• Open regulatory challenges will focus on the impacts of AI in ethics, transparency, fairness, safety, sociology and sustainability.
• Multidisciplinary work is, more than ever, of utmost importance and useful.

Structure: Risk-based (prohibited: e.g., social scoring; high-risk: e.g., AI hiring; limited/minimal: e.g., chatbots).

MSME Implications: Classify your AI (most MSMEs use minimal-risk); require transparency for limited-risk.

What You Must Do: Conduct risk assessment checklist; maintain usage logs (European Commission "AI Act" 2024).

Example: A MSME using AI for customer support must label it as AI-generated.

This risk framework is crucial for ethical deployments, like chatbots for service.

Practical Example: A retail MSME classifies its inventory AI as minimal-risk, logging usage to comply—saving time for growth-focused toolz.

Principles: Consent, minimization (only collect needed data), lawful processing.

What You Must Do: DPIA for high-risk AI; get explicit consent (European Data Protection Board "Opinion" 2024).

MSME Example: A service MSME uses GDPR-compliant AI for email personalization, boosting engagement 18%.

Practical Example: A bakery MSME conducts DPIA for AI customer loyalty apps, ensuring data minimization and gaining 12% repeat business—preparing for ethical bias checks in Unit 1.3.

Liability: MSMEs liable for AI harms (e.g., wrong recommendations); use vendor contracts with clauses.

What You Must Do: Review tools pre-deployment; document decisions (Organisation for Economic Co-operation and Development "Governing with AI" 2025).

MSME Tip: For a small manufacturer, audit AI suppliers for compliance.

Practical Example: A consulting MSME includes liability clauses in AI software contracts for client reports, preventing disputes and enabling expansion to customer service bots.

Overview:

International standard for AI management systems, focusing on governance and risk management.

What You Must Do: Establish an AI policy; audit annually; train staff on basics (International Organization for Standardization "ISO/IEC 42001" 2023).

Example: A MSME implements for AI inventory, gaining client trust. Find it in our AI digital repository!

Practical Example: A design MSME drafts a simple AI policy for tool use, auditing quarterly to certify; boosting partnerships and readiness for Module 4's loyalty programs.

Instructions:

• For your MSME, classify a potential AI tool (e.g., chatbot) under EU AI Act.
• Prompt an AI: "Assess risk level for AI customer support in a small European shop per EU AI Act."
• Add checks: Include GDPR consent steps.
• Refine: Propose a compliance matrix.

Goal: Prioritize legal attitudes.

Our understanding of what is right and wrong, as well as the rules that individuals and the society should follow.

What do we mean by “AI ethics”?

AI ethics refers to the moral principles and values that should guide the development and use of artificial intelligence systems.
AI systems should be designed and used in a way that benefits society and follows ethical principles.
Issues like fairness, transparency, privacy and safety are important ethical considerations.

AI has the potential for immense impact, both positive and negative.

AI systems are increasingly being entrusted with high-stakes decisions that affect human lives and society.

AI learns from data, which can encode human biases and prejudices. Unethical applications could automate or worsen injustice.

As AI gets more capable and autonomous, we need to maintain human accountability and control.

Ethical issues around AI use, like privacy invasion or job losses, require careful consideration to avoid public backlash.

Ethics helps ensure AI development and use is guided in a socially responsible direction.

• Responsible AI is a framework that guides how we should address the challenges around artificial intelligence from both an ethical, technical and legal point of view.
  • We must resolve ambiguity for where responsibility lies if something goes wrong!
• This framework relies on fundamental principles:
  • Accountability
  • Interpretability
  • Fairness
  • Safety
  • Privacy

• AI models learn from existing data collected from the real world, and so an accurate model may learn or even amplify problematic pre-existing biases in the data based on sensitive characteristics
• Regarding this issue, we should:
  • Interact with social scientists, and other relevant experts to understand and account for various perspectives
  • Consider how the technology and its development over time will impact different use cases (e.g., what outcomes does this technology enable)
  • Assess fairness in our datasets (e.g., identifying representation and corresponding limitations)
  • Check the system for unfair biases
  • Analyse the performance of the system, taking into account the different metrics.

• Safety and security ensure that AI systems behave as intended, regardless of how attackers try to interfere.
• Regarding this issue, we should:
  • Consider if there are incentives to make the system misbehave.
  • Identify what unintended consequences would result from the system making a mistake and assess the likelihood and severity of these consequences.
  • Build a rigorous threat model to understand all possible attack vectors.
  • Research into adversarial machine learning, as it continues to offer improved performance for defenses and provable guarantees.
  • Check if there are other vulnerabilities in the AI supply chain.

• AI systems must prioritise and safeguard consumers’ privacy and data rights and provide explicit assurances to users about how their personal data will be used and protected.
• Regarding this issue, we should:
  • Identify whether AI models can be trained without the use of sensitive data.
  • Anonymise and aggregate incoming data using best practice data-scrubbing pipelines (e.g., removing personally identifiable information (PII) and outlier or metadata values that might allow de-anonymisation).
  • Train models using federated learning, where a fleet of devices coordinates to train a shared global model from locally-stored training data.
  • Perform tests based on “exposure” measurements or membership inference assessment to estimate whether the model is unintentionally memorising or exposing sensitive data.

Automation is often seen as a threat to manual labour. Robots replace factory workers and farmers, while self-driving cars come after human drivers' jobs.

But Al is also trained to perform more intellectual tasks such as translation, programming, writing, medical diagnosis or teaching.

• Who should be blamed when decisions made by Al lead to harm?
• What are the differences between the rules that should apply to ethical Al and those that govern our own ethical behaviour?
• To what extent can we ignore the risks of Al to enjoy the benefits it provides?
• What can we do as citizens to influence how Al is used?

From algorithmic bias and privacy violations to questions of accountability and autonomy, the ethical dimensions of Al are increasingly profound and multifaceted.

The following delves into the intricate web of ethical issues surrounding artificial intelligence, shedding light on the crucial considerations that underpin the responsible development and deployment of Al technologies.

Privacy: 
Al technologies, especially in surveillance and data analysis, can infringe upon individuals' privacy rights. Striking a balance between security and personal privacy is crucial.

Accountability:
Determining who is responsible when Al systems make errors or harmful decisions is often unclear. Establishing legal and ethical frameworks for accountability is a challenge.

Transparency: 
The inner workings of some Al algorithms are opaque, making it difficult to understand why they make certain decisions. This lack of transparency can undermine trust.

Ethics in Al Research: 
Ethical considerations extend to the very research and development of Al, including issues related to data collection, experimentation, and collaboration.

Ethical Alignment:
Al researchers and engineers often face dilemmas about whether to align their work with ethical principles or prioritize commercial interests. Striking a balance can be challenging.

Intellectual Property and Open Source: 
The debate about open source versus proprietary Al technology raises ethical questions about access to and control over Al innovations.

Ethical Al use involves transparent data handling, minimizing bias, ensuring privacy, and making decisions in the best interest of society.

It requires ongoing vigilance, open communication, and a commitment to responsible Al development to maximize benefits while mitigating harm.

Effectivity: Automation, precision, speed and scalability of processes implies lower expenditure and a better prison service overall.

Prevention: AI-systems capable of detecting undue behavior and alerting officers may lower criminality, suicides, violence and smuggling of contraband.

Prediction: The power of machine learning may be harnessed to generate novel insights and predictions which may be used to decrease recidivism.

Intelligence: Data-mining may lead to knowledge that can be shared with other governmental bodies or used in united action with the police.

Justice: Humans are notoriously bad at making fair decisions. We are often prejudiced, inconsistent and often make assessments that are affected by our personal feelings. A well designed AI-system may well lead to greater justice in decision-making processes.

In this part, we'll talk about what you get and what you should leave out when it comes to working with AI.

More and more people want to use AI services such as Copilot, ChatGPT or other text-based tools in their work.

This is positive – but it is important to understand what you can and cannot do.

You may NOT paste confidential information, personal data, or sensitive information into external AI services.

This is true even if it feels like "just a draft" or if you use an AI service via a web browser on your work computer. 

Why?
AI services are often located in other countries, and we don't know exactly how data is handled or stored, for example.

The information can leave our control – and it must not happen!
 

External AI services are those that are delivered by external actors and that your company itself does not manage or have full control over. 
Examples are Chat Tools on the web or AI services from commercial platforms.

These are often located in other countries, and we don't know exactly how data is handled.

And in-house AI solutions?
These are AI systems that the company itself is responsible for, or that are delivered via the company’s own IT environments where you have control over data protection and storage.

You won't be replaced by AI.

It is you, who know your business, your colleagues and your needs – who create real benefits with AI.

Technology is a tool, but it is in combination with your experience and understanding that it becomes a powerful tool.

When used correctly, there is great potential to be able to streamline your work and free up time for more value-creating tasks.

• Here, AI can, for example, make it easier for staff to find information such as checklists and routines. 
• AI can also simplify the texts or translate them into another language. This gives staff more space to be close to the user, and at the same time reduces the risk of deviations. 
• Working with AI is also a great opportunity to explore new areas and learn more. AI can provide inspiration, new perspectives and help you quickly get an overview of topics you are curious about. 
• So use AI both as a support where you are already safe – and as a tool to grow. The best of both worlds!

Case study 1: AI in Healthcare: 

The integration of Artificial Intelligence (AI) into healthcare has brought about a transformative shift in how medical diagnosis, treatment, and patient care are approached. However, while Al offers numerous benefits, its use in these critical areas also raises significant ethical concerns that must be carefully considered to ensure that technology serves the best interests of all patients.
 

Case study 2: AI in Autonomous Vehicles:

 Autonomous vehicles (AVs) represent a ground-breaking innovation in transportation, promising to reduce traffic accidents, improve road safety, and revolutionize mobility. However, the deployment of AVS also introduces profound ethical dilemmas, particularly in scenarios where the vehicle must make decisions involving life-and-death situations. Unlike human drivers.
 

Diversity is the key.

1. Diversity of Background of the Team (to avoid "group think")

2. Diversity of Mindset (personality testing)

3. Diversity of Data

4. Diversity of Algorithmic Models

Principles: Anonymize data where possible and limit access.

MSME Examples: Encrypt and secure customer data in AI tools, such as with the use of AI chatbots.

Safeguards: Implement basic encryption and use consent forms.

Keep in mind: Secure data practices enable predictive analytics without breaches.

Communication:

Inform employees and customers about AI use (e.g., labels like "AI-generated response").

Documentation:

Keep logs of AI decisions for audits.

Accountability:

Assign roles for AI oversight in the business.

Role of Humans: AI as a tool, not replacement—e.g., using AI for routine tasks to free up time for creativity.

Ethical Decision-Making: Frameworks for when to override AI (e.g., in ethical dilemmas).

Fostering Trust: Training staff on AI ethics to build internal confidence.

MSME: A shop uses AI for stock but humans for decisions.

Practical Example: A cafe MSME trains staff to override AI suggestions in customer disputes, maintaining relationships and boosting satisfaction.

Instructions:

• Brainstorm AI bias in your MSME (e.g., customer targeting).
• Describe issue (5-10 sentences).
• Propose 3 mitigations with human oversight.
• Example: "AI ads target only young demographics—mitigate by diverse data."

Goal: Value accountability.

Practical Example: A MSME describes hiring AI bias, mitigates with diverse training data and human reviews; estimating fairer outcomes.

AI is Augmentation

It is a strategic tool that relieves teams from repetitive tasks, driving efficiency and innovation in MSMEs

Ethics is Governance

Principles like Fairness, Transparency, and Accountability are not optional; they are the core of legal compliance and trustworthy market engagement.

Regulation is Risk-Based

The EU framework (GDPR and AI Act) requires proportional compliance. MSMEs must classify their AI tools' risk level.

Key to Success

Start small, prioritize ethics! Remember: AI augments, not replaces!

Identify one low-risk, high-value and automate it with an AI tool to test the water.