WEEK 10 — DIGITAL TRANSFORMATION

Week 10: Cyber Security, Ethics and Social Issues of Digital Transformation

Agenda

  1. 1Analyze cybersecurity risks and governance frameworks in digitally transformed businesses
  2. 2Evaluate ethical implications of emerging technologies and digital business models
  3. 3Develop strategies to address social challenges arising from digital transformation
  4. 4Create balanced approaches that drive innovation while managing risks responsibly
Dr. Davood WadiDigital Transformation
Interactive four-pillar governance instrument for Week 10: selector tabs along the top switch the central framed screen between 01 Cybersecurity Vault, 02 Privacy & AI Aperture, 03 Cooling Tower Footprint, and 04 Balanced Innovation Scale. Currently active: 01 Cybersecurity & Governance01 · Cyber02 · Ethics03 · Social04 · BalanceCybersecurity & GovernancePerimeter LockedMFA + Supply AuditThreats, attack surfaces & prevention

RECAP · INDUSTRY 4.0

Review of Last Week

Interactive five-station review escapement for Week 9: Q1 First Industrial Revolution (1760–1840), Q2 Second Industrial Revolution (1870–1914), Q3 Third Industrial Revolution (1960s–Present), Q4 Fourth Industrial Revolution (Industry 4.0), and Q5 Expected Changes in Industry 4.0. Currently inspecting Q1: First Industrial RevolutionQ1 · 1760–18401st RevQ2 · 1870–19142nd RevQ3 · 1960s–Now3rd RevQ4 · Industry 4.04th RevQ5 · 4.0 Impact4.0 ShiftQ11760–1840First Industrial RevolutionSteam Power & MechanizationCoal, iron, steam engines, textile mills & centralized factories

CHAPTER 01

The Evolving Threat Landscape

CYBERSECURITY RISKS

The Evolving Threat Landscape

  • Emergence of sophisticated cyber threats (, , )
  • Attack surface expansion through IoT (e.g., Device Manipulation), cloud adoption, and remote work (e.g., Multifactor Authentication)
  • Economic impact of cybersecurity breaches on business operations and reputation
Interactive cybersecurity threat dome (SIMULATED): select State-Sponsored Attacks, Ransomware, or Supply Chain Vulnerabilities, and drag the governance & MFA defense slider (65%) to contract residual breach exposure (41%)State-SponsoredRansomwareSupply ChainSIMULATED · Breach Risk: 41%IoT DevicesDevice ManipulationCloud AdoptionShared Control PlaneRemote WorkMultifactor Auth (MFA)State-Sponsored Attacks · Cloud AdoptionTargeted intrusion into cloud infrastructure & operationsMFA & Governance: 65%
Exercise · Match up

Match each cybersecurity threat or attack-surface vector with the scenario that illustrates it.

Threat or Attack-Surface Vector

Real-World Scenario

Tap a term, then the example it goes with.

CLASS INVESTIGATION

Question

  • Research a recent cyber security breaches in North America.
  • Find out why the breach happened.
    • Who caused it?
    • What vulnerability enabled it?
  • How could the organization/business prevented the breach?
Interactive North America breach investigation board: compare three breach cases across Who caused it, What vulnerability enabled it, and How the business could have prevented it. Prevention safeguard is currently OFFCase 1: CredentialsCase 2: Supply ChainCase 3: IoT RansomSafeguard: OFF1. Who caused it?External cybercrimecredential syndicate2. What vulnerability?Reused passwords withoutMultifactor Auth (MFA)3. PreventionEnforce hardware MFA& login rate-limitingBREACH ACTIVE

INFORMATION LITERACY

Note on Research

Interactive research source verifier: inspect three breach headlines inside a digital browser screen to expose fake GenAI hallucinated breach reports versus verified regulatory records. Currently scanning: Report A: '2024 Mega-Bank Vault Breach' (HALLUCINATED REPORT)Source Viability Scanner · Click a ReportReport A: '2024 Mega-Bank Vault Breach'HALLUCINATED REPORT — Fabricated by GenAI — did not happen!Report B: '2024 National Telecom Wipeout'Click to verify source viabilityReport C: '2020 CRA & H&R Block Leak'Click to verify source viabilityHALLUCINATED REPORT · GenAI Fake: Did Not Happen!

CHAPTER 02

Case Study: Canadian Revenue Agency (CRA) Data Leak

CASE STUDY

Case Study: Canadian Revenue Agency (CRA) Data Leak

Interactive forensic diagram of the Canadian Revenue Agency (CRA) data leak across three stages: 01 H&R Block Canada data used to alter direct deposit routing and steal over $6 million; 02 Internal error exposing over 30,000 Canadian taxpayers' SINs, tax returns, and financial details; 03 Office of the Privacy Commissioner (OPC) investigation under the Privacy Act. Currently viewing Stage 0101 · H&R Block · $6M+ Stolen02 · 30,000+ SINs Exposed03 · OPC & Privacy ActH&R Block Canada (2020)ConfidentialData ObtainedUnauthorized AccessCRA Direct Deposit SwitchTaxpayerFalse Tax Returns Filed$6M+Fraudulent RefundsDirect deposit info alteredby unauthorized imposters
  • Canadian Revenue Agency (CRA)
  • Dating back to 2020
  • Hackers obtained confidential data used by H&R Block Canada, one of the country's largest tax preparation firms
  • Unauthorized access allowed imposters to change direct deposit information, submit false tax returns, and steal over $6 million in fraudulent refunds
  • CRA confirmed that an internal error led to the exposure of over 30,000 Canadian taxpayers’ personal information.
  • Resulted in Social Insurance Numbers (SIN), tax returns, and financial details being left accessible to unauthorized individuals.
  • An investigation is ongoing
  • Office of the Privacy Commissioner of Canada (OPC) is examining whether the CRA met its obligations under the Privacy Act
Exercise · Put it in order

Arrange the events of the Canadian Revenue Agency (CRA) data leak case study into their logical and chronological sequence.

Tap a step, then the number where it belongs.

CHAPTER 03

Ethics in Digital Transformation

PRIVACY BY DESIGN

Ethics in Digital Transformation

  • Privacy by design methodologies
  • Data minimization vs. data-driven business models
    • Instead of protecting the collected data, do not collect data at all.
  • Balancing personalization and privacy rights
    • Personalization needs personal data.
    • Privacy needs little data collection.
Interactive privacy-by-design iris aperture (SIMULATED): drag the data collection knob (10%) or click 'Do Not Collect Data At All' to see how Data Minimization preserves Privacy Rights (90%) while Data-Driven Personalization (10%) demands personal dataPrivacy Rights & Minimization90%Privacy Protection IndexPrivacy needs little data collection.Privacy by Design Mode:Do Not Collect Data At All (0%)SIMULATED · Collection ApertureCollecting 10% Personal DataData-Driven Personalization10%Personalization DepthPersonalization needs personal data.Data-Driven Model Mode:Collect Personal Data (85%)Minimization (0%)Max Collection (100%)
Exercise · Sort into groups

Sort each product architecture decision into the ethical data approach it represents.

      Tap a card, then tap the group it belongs to.

      RESPONSIBLE AI

      Ethical AI and Algorithm Governance

      • Algorithmic bias and fairness considerations
      • Transparency and explainability requirements
        • How is the AI model making decisions?
      Interactive AI algorithm governance X-ray: select Social bias, Racial bias, Gender bias, or Cognitive bias, and toggle the Transparency & Explainability X-ray lens (OPEN) to inspect how the AI model is making decisionsSocial biasRacial biasGender biasCognitive biasHow is the AI model making decisions? · Explainable ViewSocioeconomic Proxy Weight: 78%Skewed against lower-income tiersFix: Audit & strip socioeconomic proxiesTransparency & Explainability X-Ray: ON (Click to Hide)
      Exercise · Which one is it?

      Identify which algorithmic bias or governance failure is illustrated by each AI scenario.

      A loan-scoring model penalizes applicants from lower-income zip codes and non-elite schools, reproducing socioeconomic class stratification.

      Which concept does this case show?

      INFORMATION INTEGRITY

      Truth, Trust and Digital Information Integrity

      • Corporate responsibility in content moderation
        • The balance between battling misinformation and censorship
      Interactive content moderation balance fulcrum (SIMULATED): drag the moderation strictness knob (50%) between Battling Misinformation and Censorship Risk. Current state: Balanced Governance: Battles Misinformation Without CensorshipSIMULATED · Balanced Governance: Battles Misinformation Without CensorshipMisinformation & Synthetic MediaOverreaching CensorshipMisinformation Risk (0%)Censorship Risk (100%)

      CHAPTER 04

      Environmental Sustainability & Platform Power

      CARBON FOOTPRINT OF AI

      Environmental Sustainability

      • Carbon footprint of digital technologies
      • GPT-3 (2020): Estimated training cost ranged from $2 million to $4.6 million.
      • GPT-4 (2023): Training costs were estimated between $41 million and $78.4 million. OpenAI CEO Sam Altman suggested the total cost exceeded $100 million when including staff salaries.
      • Gemini Ultra (2023): Google's advanced AI model had estimated training costs between $30 million and $191 million, not including staff salaries.
      Interactive frontier AI training cost and carbon footprint comparator: GPT-3 (2020, $2M–$4.6M), GPT-4 (2023, $41M–$78.4M, exceeding $100M with staff salaries), and Gemini Ultra (2023, $30M–$191M excluding staff salaries). Currently inspecting: GPT-4 (2023)GPT-3 (2020)GPT-4 (2023)Gemini Ultra (2023)+ Staff Salaries: OFF (Click)Carbon & Compute ScaleGPT-4 (2023) FootprintEstimated Training Cost Range ($ Millions USD)GPT-3 (2020)$2M–$4.6MGPT-4 (2023)$41M–$78.4MGemini Ultra (2023)$30M–$191MGPT-4 (2023): $41 million to $78.4 million (> $100M with staff salaries)

      EXPONENTIAL TRAJECTORY

      Environmental Sustainability

      Interactive 2017–2027 AI training cost growth instrument: from $930 for the 2017 Transformer model to over $78 million for GPT-4 in 2023, compounding at 2.5x per year (between 2x and 3x per year) to exceed $1 billion (~$3.06B) by 20272017–2027 Frontier Training Cost Trajectory (Log Scale · 2.5x / Year)$9302017 · Transformer$2M–$4.6M2020 · GPT-3Over $78M2023 · GPT-4~$3.06B (>$1B)2027 · ProjectionGrowth Factor: 2.5x / yr (2x–3x)2027: >$1B (~$3.06B)
      • From 2017 to 2023, costs rose from $930 for the Transformer model to over $78 million for GPT-4.
      • Training costs for frontier AI models have grown by a factor of 2x to 3x per year for the past eight years.
      • If this trend continues, the largest training runs could cost more than a billion dollars by 2027.
      Knowledge Check

      Between 2017 and 2023, frontier AI training costs rose from $930 for the original Transformer model to over $78 million for GPT-4, growing by a factor of 2x to 3x per year. If an AI lab's flagship training run cost $100 million in 2023 and follows the upper 3x annual growth rate, what does this trajectory imply for 2027?

      MARKET CONCENTRATION

      Platform Power and Digital Monopolies

      • Network effects and winner-takes-all dynamics
        • In 2024, if you were to host your website, you had to go through one of top five cloud companies.
      Interactive platform power and cloud monopoly funnel (SIMULATED): drag the network effects slider (90%) to watch independent websites converge through the top five cloud companies in 2024 under winner-takes-all dynamicsIndependent Websites & Digital Businesses Seeking Hosting2024 Winner-Takes-All Gateway: Top Five Cloud CompaniesCloud #1Cloud #2Cloud #3Cloud #4Cloud #5SIMULATED · Network Effects: 90%
      Knowledge Check

      In 2024, a startup launching a new web application discovers that to host its site reliably at scale, it effectively must go through one of the top five cloud companies. Which economic mechanism from the lecture explains this concentration of infrastructure power?

      RECOMMENDED LITERATURE

      Further Readings

      • O'Neil, C. (2016). Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy
      Interactive three-gear diagram of Cathy O'Neil's Weapons of Math Destruction (2016): 01 Opacity, 02 Scale, and 03 Inequality & Democratic Harm. Currently inspecting Gear 01: Opacity (Black-Box Scoring)O'Neil (2016) · How Big Data Increases Inequality and Threatens Democracy (Click a Gear)01 · Opacity02 · Scale03 · Damage01 · Opacity (Black-Box Scoring)Citizens and workers are scored by proprietary mathematical models they cannot inspect or appeal.

      CONCLUSION

      Conclusion

      KEY TAKEAWAYS

      1. 1Digital Transformation Cybersecurity Threats
      2. 2Responsible AI and AI Ethics
      3. 3Environmental Impacts of Artificial Intelligence
      4. 4Centralization of Cloud Computing Power