Welcome and Introduction

Setting the Stage

Rainer Michaeli, Director of the Institute for Competitive Intelligence, welcomes participants and introduces Edwin Vlems, a colleague he has known for approximately 20 years. Their professional relationship began at a WIN conference focused on marketing intelligence, where Vlems had already developed advanced systems for automated newsletter creation.

Michaeli emphasizes that while everyone has experimented with AI, the critical questions remain: What comes next? How does regulation intersect with AI adoption? The European Union's regulatory framework provides both challenges and opportunities for competitive intelligence professionals.

Session Format

• The webinar is recorded for later distribution
• Participants use chat for questions during the presentation
• Q&A session follows the main presentation
• Cameras and microphones remain off during the session

Background: The Personality Economy

Edwin Vlems: Professional Background

Edwin Vlems is a speaker and teacher on marketing intelligence and AI, founder of ComaxxAI with focus on AI literacy. His AI journey began 25 years ago during a previous "AI summer" when he and his brother built AI applications. The field has experienced cyclical patterns since the 1950s—periods of intense interest ("AI summers") followed by declining enthusiasm ("AI winters").

Following an AI winter and European recession in the early 2000s, Vlems authored five marketing books. In 2022, tools like Jasper (which could generate articles from titles alone) reignited his interest, months before ChatGPT's public launch.

The Personality Economy Concept

Vlems wrote his sixth book in two hours using ChatGPT, inspired by a TikTok video suggesting that AI would eliminate the knowledge economy. The central thesis: people will be hired not for knowledge (now accessible through AI) but for personality shaped by years of experience.

The book was published through Amazon's self-publishing platform, selling approximately one copy per week—an interesting experiment in AI-assisted content creation rather than a commercial venture.

Overview of the EU AI Act

Europe's Regulatory Approach

Europeans often resent Brussels regulations, but Vlems argues the EU AI Act deserves pride of place. As global AI adoption intensifies, Europe will have established protective rules against excessive dependence and manipulation.

The Dual Nature of the Act

The AI Act represents Europe's first regulation balancing risks and opportunities—the "yin and yang" of AI governance. Unlike previous regulations focusing exclusively on limitations, this framework also encourages understanding AI's full potential for innovation and growth.

Session Agenda

• Introduction to AI literacy in Europe
• Deep dive into the AI Act's structure and requirements
• Practical applications for improving competitive intelligence

Key Dates for the EU AI Act

Implementation Timeline

The AI Act follows a phased rollout approach with four critical milestones:

February 2, 2025 (Already in Effect)

• Prohibitions on unacceptable AI practices now enforced
• AI literacy requirements mandatory for all deployers
• Many professionals currently in violation without formal AI literacy training

August 2, 2025

• General-purpose AI model obligations begin
• Applies to most companies and sectors using broad AI applications

August 2, 2026

• Full enforcement for high-risk AI systems
• Covers medical devices, critical infrastructure, HR systems

August 2, 2027

• Complete implementation of all provisions scheduled
• Legal experts acknowledge uncertainty about meeting this deadline

The AI Act: The First Articles

Foundation Through First Principles

The Act's first four articles establish the regulatory foundation. Vlems uses the yin-yang metaphor to explain the unique dual approach: protecting against risks while promoting innovation.

The Yin: Protection and Safety

• Guarding health, safety, and fundamental rights
• Ensuring democracy, rule of law, and environmental protection
• Establishing clear boundaries for unacceptable practices

The Yang: Innovation and Growth

• Creating a unified European internal market
• Supporting safe technology development
• Promoting AI literacy for competitive advantage

This represents the first continental regulation that actively encourages understanding both AI's potential benefits and necessary safeguards.

Article 1: Purpose of the AI Act

Human-Centric and Trustworthy AI

The primary objective promotes AI adoption that is reliable and respects human values established in European charters. This creates a foundation for AI systems that serve people rather than manipulate them.

High Levels of Protection

The Act ensures protection for:

• Health and safety of EU citizens
• Fundamental rights and democratic values
• Rule of law across member states
• Environmental sustainability

Unified European Market

Harmonizing rules across the EU improves market functioning and enables effective cross-border cooperation. This prevents regulatory fragmentation while maintaining high standards.

Supporting Innovation

The Act creates a legal environment where European businesses can safely lead in technology development, balancing protection with progress.

Article 2: Application

Who Must Comply?

The Act applies to three primary categories:

EU and Global Providers

• Anyone placing AI systems on the EU market, regardless of location
• Must maintain an office in Europe contactable by authorities
• Includes companies like OpenAI and Microsoft operating in Europe

EU-Based Deployers

• Organizations using AI systems under their authority within the EU
• Must ensure staff possess AI literacy
• Responsible for appropriate use policies and oversight

International Output Usage

• Third-country providers whose AI output is used inside the EU
• Even systems operating outside Europe fall under the Act if outputs affect EU citizens

Exemptions

• Defense and National Security: Separate frameworks govern military applications
• Scientific R&D: Research systems before market entry are exempt
• Non-Professional Use: Personal, non-commercial AI use is not regulated

Geographic Scope

The Act covers the European Union plus EEA countries (Iceland, Norway, Liechtenstein). The United Kingdom has developed separate post-Brexit regulation.

Article 4: AI Literacy

Definition and Scope

AI literacy encompasses the skills, knowledge, and understanding necessary to make informed decisions about AI, including awareness of risks and potential harms. This goes beyond technical knowledge to include ethical, legal, and practical considerations.

Mandatory Requirements

Organizations must ensure a "sufficient level" of literacy for:

• All staff operating AI systems on their behalf
• Decision-makers implementing AI strategies
• Personnel whose work is affected by AI outputs

Customized Training Approach

No one-size-fits-all solution exists. Training must be tailored based on:

• Technical expertise of personnel
• Specific roles and responsibilities
• Context of AI use within the organization
• Risk level of systems being deployed

For example, HR departments using AI for hiring decisions require different training than marketing teams using AI for content generation.

Active Since February 2025

The literacy mandate took effect early—one of the first requirements enforced. This emphasizes that compliance and risk mitigation apply across all AI risk levels, not just high-risk systems.

Strategic Priority vs. Compliance Checkbox

Beyond avoiding penalties, AI literacy represents a strategic necessity. It unlocks innovation potential, protects fundamental rights, and enables employees to work effectively with AI tools—particularly valuable in competitive intelligence applications.

Unacceptable AI Risk

Six Prohibited Practices

The following AI applications are completely banned in the European Union:

1. Cognitive Manipulation

• Systems designed to create emotional dependency
• AI that encourages users to "fall in love" with the system
• Prohibited for both adults and children
• Meta recently disclosed such systems—illegal under EU law

2. Social Scoring

• Systems like China's social credit mechanisms
• Linking unrelated behaviors (e.g., traffic violations preventing vacation travel)
• Digital scoring systems that restrict citizens' rights

3. Real-Time Remote Biometrics in Public Spaces

• Commercial use of facial recognition cameras
• Private deployment of identification systems
• Different rules apply to law enforcement with strict oversight

4. Biometric Categorization

• Classifying people by sex, race, or religious beliefs
• Automated sorting based on protected characteristics

5. Emotion Recognition at Work and School

• Despite AI's capability to detect emotions accurately
• Workplace monitoring of employee emotions is forbidden
• Educational institutions cannot deploy emotion-reading systems on students

6. Predictive Policing

• Systems predicting future criminal behavior
• Prohibited even for law enforcement agencies
• Despite Hollywood depictions, such systems cannot legally operate in Europe

High-Risk AI Systems

Two Categories Defined by Annexes

High-risk classification does not mean prohibition—it requires compliance with extensive requirements and oversight.

Category A: Products (Annex I)

AI used as safety components in regulated products:

• Medical devices and diagnostic systems
• Machinery and industrial equipment
• Toys with AI functionality
• Automobiles and transportation systems
• Elevators and lifts
• Aviation security systems

When AI is integrated into these products, manufacturers must comply with high-risk regulations, often requiring legal expertise. For consumers, this ensures AI-equipped elevators and medical devices meet rigorous EU safety standards.

Category B: Critical Areas (Annex III)

• Biometrics: Identification and verification systems
• Critical Infrastructure: Energy, transportation, water supply
• Education and Employment: Admission decisions, hiring, performance evaluation
• Essential Services: Credit scoring, insurance underwriting
• Law Enforcement: Crime analysis, evidence evaluation
• Migration and Border Control: Visa decisions, asylum applications
• Justice and Democracy: Legal research, electoral processes

The Decision Threshold

The critical factor is whether AI makes decisions or merely provides advice:

• High-risk: AI autonomously decides who receives credit, employment, or insurance
• Lower risk: AI provides recommendations that humans evaluate and decide upon

Limited Risk: Transparency

Most Companies Fall Here

The limited-risk category focuses on transparency requirements rather than extensive oversight. Four key obligations apply:

1. Chatbot Disclosure

• Users must know they are communicating with a machine, not a person
• Clear identification required at the start of interaction
• Applies to customer service bots, support systems, and conversational AI

2. Deepfake Labeling

• Artificially generated content appearing real must be clearly marked
• Particularly important for synthetic images, videos, and audio
• Labels must be visible and unambiguous

3. Watermarking for Detection

• Machine-readable metadata embedded in AI-generated content
• Enables other AI systems to identify synthetic material
• Technical standard for content provenance

4. AI Literacy Obligation

• Staff and users must be competent in AI system use
• Organizational responsibility for training
• Applies even to limited-risk systems

Transparency Drives Credibility

The shift from "black-box" predictions to auditable, documented decision trails improves stakeholder trust. This transparency requirement aligns with competitive intelligence best practices: verify everything, trust nothing without corroboration.

The Cost of Non-Compliance

Penalties Mirror GDPR Structure

Violation fines are calculated as the higher of a fixed amount or percentage of global annual turnover:

Prohibited AI Violations

• €35 million OR 7% of global annual turnover
• For large companies, 7% can lead to bankruptcy
• Applies to any use of the six forbidden practices

High-Risk and General-Purpose AI Violations

• €15 million OR 3% of global annual turnover
• Covers improper deployment of high-risk systems
• Includes failure to meet documentation and oversight requirements

Misinformation and Inadequate Disclosure

• €7.5 million OR 1.5% of global annual turnover
• Penalties for failing to label AI-generated content
• Applies to transparency requirement violations

Ignorance Is Not a Defense

Organizations cannot claim they were unaware whether their AI was prohibited or high-risk. Lack of awareness itself constitutes a violation subject to penalties. This creates strong incentives for proper AI literacy and classification assessments.

The 7 Requirements of Trustworthy AI

Comprehensive Framework for AI Deployment

Seven interconnected requirements define what makes AI systems trustworthy under EU law:

1. Human Agency and Oversight

• Always maintain "human in the loop"
• People must oversee AI operations
• Authority to intervene when necessary
• Prevents fully autonomous decision-making in critical contexts

2. Technical Robustness and Safety

• Protection against hacking and exploitation
• Security standards applicable to all EU software
• Reliability under varied operating conditions

3. Privacy and Data Governance

• Significant overlap with GDPR requirements
• Uploading personal information to AI systems is forbidden
• Critical: The upload itself constitutes a breach, even without data leakage
• Data governance regulations currently under development

4. Transparency

• Disclosure of AI use to customers, government, and affected parties
• If AI is used in customer-facing work, disclosure is mandatory
• Explainable decision-making processes

5. Diversity, Non-Discrimination, and Fairness

• Chatbots often trained on biased historical data
• Organizations must ensure AI does not discriminate by gender, race, or protected characteristics
• Mistral is currently the only major chatbot fully compliant with this requirement
• As Europe's primary AI provider, Mistral operates all servers within the EU
• Server location prevents external surveillance (e.g., by CIA)

6. Societal and Environmental Well-Being

• Consideration of AI's energy consumption
• Particularly important as global usage scales
• Assessment of broader societal impacts

7. Accountability

• Demonstrate AI usage to government authorities
• Conduct regular assessments
• Document staff training programs
• Maintain audit trails for review

Better CI with the AI Act

Regulation as Quality Framework

The core message: properly implemented, the AI Act automatically improves competitive intelligence quality. The regulation should not be viewed as purely restrictive—when done well, it enhances CI effectiveness.

The LinkedIn Profile Data Breach Example

A frequent question illustrates the intersection of AI regulation and CI ethics: Can LinkedIn profiles of competitor employees be uploaded to AI for competitive research?

While this seems logical—profiles appear public and accessible—EU law clearly prohibits it. Four reasons explain why this constitutes a data breach:

• Protected Personal Data: Despite visibility on LinkedIn, profiles remain protected personal data requiring permission for AI processing
• Platform Status: LinkedIn is not technically open; detailed access requires membership, creating privacy expectations
• Active Processing: AI chatbots are data processors; uploading profiles constitutes processing under GDPR
• Regulatory Overlap: The AI Act does not replace GDPR but works alongside it

Elite CI Behaviors Formalized

Most professionals aware of quality CI practice already know that uploading personal information is inappropriate. The AI Act formalizes what elite practitioners already do, transforming best practices into legal requirements.

For organizations seeking to strengthen their competitive intelligence capabilities while maintaining compliance, the Certificate in Competitive Intelligence Research (CCIR) provides comprehensive training in ethical research methodologies and data governance.

Why Literacy Improves CI

Three Dimensions of Improvement

1. Fewer False Signals

• Teams trained to question AI outputs spot errors early
• Critical evaluation becomes systematic rather than occasional
• Reduces propagation of AI hallucinations into strategic decisions

2. Nuanced Communication

• Accurately convey uncertainty and confidence levels
• AI can hallucinate—literacy training teaches recognition
• Intelligence derived from AI requires appropriate qualification
• Stakeholders receive context about data sources and reliability

3. Innovation Without Recklessness

• Confident tool use promotes efficiency gains
• Understanding boundaries prevents inappropriate applications
• Good CI professionals avoid recklessness with or without AI
• The Act formalizes this principle into systematic practice

AI Literacy as Competitive Advantage

Organizations implementing comprehensive AI literacy programs gain advantages beyond compliance. The Fundamental Certificate in Competitive Intelligence (FCCI) includes AI literacy components specifically designed for market and competitive analysis professionals, transforming regulatory requirements into skill development opportunities.

Professionals seeking advanced capabilities in AI-assisted analysis can explore the complete workshop catalog, which integrates AI best practices throughout the curriculum.

Verification & Source Triangulation

The Law (Recital 20)

Users must evaluate AI functionality and not blindly trust outputs. This creates a legal requirement for critical assessment rather than passive acceptance of AI-generated intelligence.

The CI Practice: The Two-Person Rule

Quality competitive intelligence requires at minimum four eyes on any analysis:

• Cross-checking AI insights against primary sources
• Independent verification by a second analyst
• Zero reliance on a single source of truth
• Documentation of verification steps

Practical Implementation

• AI output serves as initial hypothesis, not final conclusion
• Every claim requires corroboration from independent sources
• Conflicting information triggers deeper investigation
• Source diversity strengthens analytical confidence

Professional Development

The Kompetenz-Training für professionelle Marktrecherchen und -Analysen workshop provides systematic training in verification methodologies and source triangulation techniques that satisfy both AI Act requirements and professional standards.

Mitigating Blind Spots

The Law's Requirement

Deployers must understand training data limitations and potential system-related weaknesses. Employers must provide training about these limitations to staff using AI systems.

The CI Practice: Detecting Skewed Market Signals

AI training data reflects historical patterns and geographic biases. If a region, industry segment, or time period is underrepresented, AI outputs will reflect this gap:

• Emerging markets may have less training data than established ones
• Recent developments post-training may be completely absent
• Niche industries receive less coverage than mainstream sectors
• Language barriers create information asymmetries

Compensation Strategies

• Supplement AI analysis with targeted primary research
• Consult regional experts for underrepresented markets
• Use multiple AI systems with different training approaches
• Maintain awareness of knowledge cutoff dates

Most experienced CI professionals already employ these methods without AI Act mandates. The regulation formalizes professional practice into legal requirements.

The Fallstricke und kognitive Wahrnehmungsfehler workshop addresses cognitive biases and analytical pitfalls, including those introduced or amplified by AI systems.

Transparency & Traceability

The Law's Requirements

• Automatic logging of AI system events
• Explainable decisions—no "black box" outputs
• Auditability of AI-assisted processes
• Documentation of AI involvement in deliverables

The CI Practice: Robust Audit Trails

Professional competitive intelligence maintains comprehensive documentation:

• Prompts: Record exact queries submitted to AI systems
• Inputs: Document source materials and data provided
• Outputs: Preserve AI-generated content before human editing
• Rationale: Explain why AI was used and how outputs were validated
• Modifications: Track changes made to AI-generated content

Defending Conclusions

If analysis is challenged by stakeholders:

• Show the complete analytical process
• Demonstrate verification steps taken
• Explain how AI outputs were integrated with human expertise
• Provide access to source materials

The law and elite CI practice operate in parallel—both demand transparent, traceable processes that build stakeholder confidence.

Organizations building or refining CI capabilities benefit from the Reporting und Kommunikation workshop, which addresses transparency standards and documentation requirements for AI-enhanced intelligence work.

The Human in the Loop

The Law's Mandate

Natural persons must oversee AI systems and possess authority to intervene. This prevents fully autonomous decision-making in contexts affecting people's lives and rights.

The CI Practice: Scenario Thinking

Quality analysts ask: "What if the model is wrong?"

• Develop alternative scenarios challenging AI conclusions
• Identify assumptions embedded in AI outputs
• Test robustness of findings against contradictory evidence
• Apply domain expertise AI lacks

Adding Human Context

AI systems cannot:

• Understand organizational culture and politics
• Grasp strategic implications beyond pattern recognition
• Apply judgment honed through years of industry experience
• Recognize when "correct" answers miss strategic reality

Good analysts add this context that AI might miss. In quality CI practice, work cannot be left solely to AI tools—human judgment and oversight remain essential.

The Kreatives und kritisches Denken workshop develops the analytical capabilities that complement AI systems, ensuring human expertise guides technological tools rather than deferring to them.

Implementation: Governance

Establishing Acceptable Use Policies

Organizations must create clear guidelines governing AI use in competitive intelligence:

1. No Sensitive Data in Public Prompts

• Personal data: No uploading of European citizens' information without consent
• Commercial data: No proprietary company information in public AI systems
• Competitor data: No LinkedIn profiles, personnel information, or protected materials

2. Mandatory Labeling of AI-Generated Content

• Mark AI-produced sections of reports and presentations
• Distinguish AI-assisted from human-created analysis
• Particularly important when content is not easily identifiable as synthetic
• Clear labeling where AI made recommendations later adopted

3. Verification Checkpoints for AI Facts

• Systematic fact-checking before distribution
• Source verification for all quantitative claims
• Cross-referencing AI-discovered information
• Documentation of verification process

Building Governance Infrastructure

Organizations establishing or refining competitive intelligence functions benefit from structured guidance. The Inhouse-CI-Center aufbauen und betreiben workshop addresses governance structures, policy development, and operational frameworks that integrate AI Act compliance with CI effectiveness.

For executives overseeing intelligence operations, the Strategic Competitive Intelligence for Executives (SCIE) program provides leadership-level perspective on balancing innovation, compliance, and strategic value.

The AI-Literate Analyst

Four Defining Characteristics

1. Skeptical (Verifies Outputs)

• Treats AI outputs as hypotheses requiring validation
• Questions rather than accepts AI-generated conclusions
• Maintains healthy professional skepticism
• This skepticism characterizes quality CI professionals regardless of AI use

2. Transparent (Documents Steps)

• Creates clear audit trails for all analytical work
• Documents AI involvement in research and analysis
• Makes methodology visible to stakeholders
• Enables review and validation by colleagues

3. Aware (Understands Bias)

• Recognizes training data limitations
• Identifies potential bias in AI outputs
• Understands what works and what does not in AI applications
• Spots potential misinformation sources

4. Strategic (Human Judgment)

• Applies higher-level thinking to AI outputs
• Integrates AI insights with strategic context
• Exercises judgment AI cannot replicate
• Maintains ultimate responsibility for conclusions

Developing AI Literacy

The ICI certificate programs integrate AI literacy throughout their curricula, ensuring graduates possess both traditional CI competencies and modern AI capabilities. This integrated approach transforms regulatory requirements into professional development opportunities.

Professionals can explore upcoming workshop dates to begin building AI-literate analytical capabilities.

Turning Regulation into Value

The Framework for Trust

The EU AI Act provides a framework for building trustworthy, effective AI-enhanced competitive intelligence. By mastering this framework, professionals do not merely stay compliant—they deliver sharper, safer, and more strategic intelligence.

Two Sides of the Same Coin

The regulation and quality CI practice represent complementary forces:

• Verification requirements formalize the two-person rule
• Transparency mandates enforce audit trail best practices
• Literacy obligations systematize professional development
• Human oversight rules protect against over-automation

If the AI Act is understood and implemented properly, no changes to existing quality CI work processes are necessary. The regulation codifies what elite practitioners already do.

Strategic Advantage Through Compliance

Organizations that view the AI Act as opportunity rather than burden gain competitive advantages:

• Enhanced analytical quality through systematic verification
• Increased stakeholder trust through transparency
• Reduced risk of strategic errors from AI hallucinations
• Professional development that drives both compliance and capability

Looking Ahead

In a few years, when global AI adoption intensifies, Europe will benefit from having established protective frameworks. While other regions may struggle with AI addiction and manipulation, European professionals will operate within structures that balance innovation with responsibility.

Next Steps for CI Professionals

The Institute for Competitive Intelligence offers comprehensive programs integrating AI literacy with competitive intelligence excellence. Whether you are beginning your CI journey or advancing existing capabilities, our certificate programs and workshops provide the foundation for AI-literate, regulation-compliant intelligence work.

Questions about which program fits your needs? Contact our advisory team to discuss your professional development goals.