UPI Study › Courses › Ethics in Technology › Free lessons › Lesson 6.1 Free with a free account
Why Collecting and Analyzing Massive Amounts of Data Raises Serious Privacy Questions
Ethics in Technology · Chapter 6: When Big Data Becomes a Big Risk: Security Gaps, Legal Frameworks, and National DefenseSection 6.1 · lesson 22 of 11616 min · free with a free account
Free preview
Create a free account to keep watching
It takes a minute and costs nothing. Your free account unlocks all of Chapter 1 and 3 more lessons of Ethics in Technology.
Lesson complete
Earn the 3 college credits for Ethics in Technology
Enrolling adds a quiz after each lesson, the graded work and the official transcript. Recommended for credit by NCCRS.
Next lesson: How Large-Scale Data Storage and Processing Create New Targets for Cyber Attackers →← Applying Data Mining Techniques to Understand Behavior and Trends on Social PlatformsNext lesson: How Large-Scale Data Storage and Processing Create New Targets for Cyber Attackers →
In this lesson
Linkage, inference, purpose creep, and consequences for people.
- Explain re-identification and its limits
- Compare privacy techniques and consent weaknesses
- Recognize profiling and discriminatory outcomes
- Analyze the Cambridge Analytica case
Free study notes
Get this lesson's notes by email
The key points and 4 practice questions with answers, in your inbox. One email, free.
One email. Unsubscribe in one click.
Transcript
Everything said in the video, section by section (1,902 words). Read along, or search it.
Chapter 6
Why Big Data Raises Privacy Questions
Large datasets can improve research and services, while creating serious questions about personal privacy. A single purchase or location point may seem ordinary, but combinations can reveal a detailed history. The source groups these concerns as linkage, inference, and the persistence of information at scale. We examine why removing a name does not automatically make a released dataset anonymous. Ordinary characteristics and behavioral traces can sometimes reconnect records with the people behind them. Several protection techniques reduce risks, while introducing assumptions and tradeoffs that deserve review. We also examine why notice and consent can provide less control than a checkbox suggests. Purpose creep describes information moving into uses beyond the reason for its original collection. Profiling can infer sensitive circumstances, and automated systems may produce discriminatory consequences. The Cambridge Analytica case connects these issues, through app access, political profiling, and weak accountability. Our aim is to assess privacy claims carefully, rather than treating technology or consent as a complete solution.
Three Pressures
Combination Changes What a Record Can Reveal
Linkage occurs when separate records are joined, allowing information to be connected across different contexts. An apparently anonymous health record can become identifying, when matched with a named public list. Inference goes further, estimating a circumstance that a person never explicitly chose to disclose. Purchases may suggest health circumstances, while patterns of movement can reveal routines and relationships. An estimate can still cause harm, even when it turns out to be inaccurate. Scale and persistence increase opportunities for reuse, across years and different organizations. Information collected by one application may later reach brokers, or support decisions in another setting. These concerns also existed in smaller systems, so large-scale analysis amplifies rather than invents them. Greater detail and variety provide more opportunities to link records and derive sensitive conclusions. Privacy therefore includes the uses and consequences of information, rather than only secrecy of a name. Review what is linked, inferred, retained, and used, before accepting a claim that the data is harmless.
Re-identification
Removing Direct Identifiers Does Not Establish Anonymity
Direct identifiers include names, telephone numbers, and account information that can identify a person directly. Removing them can reduce exposure, without establishing that identification is no longer possible. Quasi-identifiers are ordinary characteristics that may distinguish someone when combined with other details. ZIP code, birth date, and sex provide the source's familiar example of such a combination. An external voter list may contain matching characteristics, alongside the person's name. Linking those sources can associate the released record, with the person who matches that combination. Uniqueness within the relevant population can support identification, when suitable outside information is available. It does not mean every unique record has already been matched, or every attempted linkage will succeed. Risk depends on detail, population size, outside sources, and the release's other protections. Pseudonymous account numbers do not prevent identification, when the remaining behavior itself contains useful clues. Anonymity requires assessment of the whole release, rather than a checklist confirming that names were removed.
Historical Evidence
Different Kinds of Detail Can Support Identification
Latanya Sweeney's research illustrated the identifying power of ordinary demographic combinations. Her two thousand publication estimated eighty-seven percent uniqueness, using historical United States census information. That figure is a study-specific historical estimate, rather than today's universal rate of successful identification. The Massachusetts example linked a health record to Governor William Weld through a voter list. The source also describes the two thousand six release of AOL search records under numbered accounts. Search behavior contained clues that journalists used to identify one of the people represented. Netflix research showed that ratings and dates, could support linkage with public movie reviews. A two thousand thirteen mobility study examined records for one point five million individuals. Four time-and-place points uniquely distinguished ninety-five percent, within that particular study dataset. This illustrates location sensitivity, without proving that all location releases are equally identifiable. Different data types share the same lesson that useful detail can also provide clues about identity.
Privacy Techniques
Protection Requires Assumptions and Tradeoffs
Pseudonymization replaces direct identifiers with codes, while preserving records for some necessary analytical purposes. Additional information and remaining patterns may still permit linkage, so the records require protection. Generalization replaces exact values with broader categories, such as age bands or larger regions. That can reduce distinguishing detail, while also reducing precision for some kinds of analysis. K-anonymity groups records with shared quasi-identifiers, so a combination matches at least the specified number. It can still reveal sensitive information, when everyone in the matching group shares that attribute. Differential privacy formally bounds how participation, can affect the distribution of released results. Calibrated random noise is a common mechanism, with privacy and accuracy affected by its settings. Implementation and repeated releases matter, rather than merely claiming that some noise was added. The Census Bureau used differential privacy in its twenty twenty Census disclosure-avoidance system. Protecting published results does not authorize exposing the confidential underlying database or every individual record.
Notice and Consent
A Checkbox Can Conceal Weak Understanding or Choice
Notice and consent attempts to disclose data practices, and give people a choice about them. Long and complex policies can make that choice difficult, even when the notice is available. The source's claim that nobody reads policies is too absolute, but limited understanding remains a concern. Refusal can carry substantial costs, when school, work, or essential services require a particular platform. That dependence raises questions about whether agreement is meaningfully free rather than practically compulsory. Future analysis may discover sensitive inferences that a person could not anticipate from the original collection. Sharing contacts, photographs, or genetic information can also reveal something, about friends and relatives. One person's choice therefore does not settle, the privacy interests of every affected person. Under the GDPR, consent is one lawful basis, rather than the only possible basis for processing. Where consent is used, its requirements need to be met; merely using a service is not universally sufficient. Notice and choice should sit alongside information limits, accountability, and safeguards for affected people.
Purpose Creep
A New Use Can Cross the Original Boundary
Purpose creep, also called function creep, moves information beyond the reason for its original collection. Each extension can appear small, while the eventual use differs substantially from what people expected. The source gives Singapore's TraceTogether program, as a historical example of damaged expectations. In January twenty twenty-one, the government acknowledged police access to contact-tracing information for investigations. Legislation then restricted criminal use to specified categories of serious offenses. The lesson concerns that historical change, rather than a claim about the program's current operation. GDPR purpose limitation restricts incompatible further use, within the regulation's applicable rules. Data minimization requires personal information to be adequate, relevant, and limited to what the purpose needs. Consent, where relied on, must be freely given, specific, informed, and unambiguous. A changed purpose needs assessment, instead of assuming that an old checkbox covers every later application. Those boundaries help preserve trust, while requiring organizations to justify what they actually do with information.
Profiling
A Pattern Can Infer a Fact That Was Never Supplied
Profiling uses information to evaluate or predict, characteristics or behavior associated with a person. Interests, reliability, health, and likely actions, can become estimates used within a personal profile. Recommendations and fraud detection illustrate purposes, where profiling may provide a useful service. Hiring, lending, insurance, and risk assessments can instead change opportunities, prices, or treatment. The source describes the reported Target pregnancy example, involving patterns in customer purchases. Shopping records were used to estimate a sensitive circumstance, rather than merely record the bought items. The privacy issue concerns inference beyond what the person supplied, and the use of that estimate. An estimate can be wrong, while still affecting the person through inappropriate contact or decisions. Correctness alone does not establish permission for every way that sensitive information could be used. Invisible profiling can leave someone unsure, why an offer, opportunity, or service was withheld. Ask how affected people learn about the process, and whether they can correct or challenge its consequences.
Discrimination
Removing a Protected Field Does Not Remove Every Risk
Automated systems can produce unfair outcomes, without a designer deliberately intending discrimination. Historical training data may preserve patterns from earlier decisions or unequal opportunities. The source's reported Amazon recruiting experiment illustrates, learning from male-dominated historical applications. It is a historical example, rather than a claim about every current hiring system. Removing race or sex does not remove correlated information from every other field. ZIP codes, schools, or activities can act as proxies, carrying information about protected characteristics. The target being predicted also matters, because an easy measurement can differ from the real need. A two thousand nineteen health-care study found racial bias, when cost stood in for health needs. The COMPAS debate concerned unequal errors, alongside disputed findings and competing ways of measuring fairness. These mechanisms require evaluation in context, rather than assuming that all algorithms inevitably behave alike. Affected people need visibility and a meaningful challenge, when a system changes their treatment or opportunities.
Cambridge Analytica
App Access Expanded Far Beyond the App’s Own Users
In two thousand fourteen, Aleksandr Kogan's personality app gathered Facebook user and friend information. Personality scores from that information supported, Cambridge Analytica's voter profiling and targeted-advertising services. The source contrasts roughly two hundred seventy thousand app users, with a much larger affected population. Facebook estimated that information on up to eighty-seven million people, may have been improperly shared. Different investigations reported different totals, reflecting their scope and the information they considered. The important relationship is the expansion from app participants into a much larger friend network. Facebook announced restrictions on friend-data access, with a transition period for existing applications. It learned of sharing in two thousand fifteen, and requested deletion of the information. Later disclosures raised questions about verification, enforcement, and what affected people had been told. Public reporting and whistleblower accounts in March two thousand eighteen, brought wider attention to the case. The privacy failures do not prove how much profiling changed voters' behavior or any election result.
Accountability
The Consequences Included Penalties and New Oversight
Cambridge Analytica shut down in two thousand eighteen, while Facebook leadership faced congressional scrutiny. The United Kingdom's regulator imposed a penalty of five hundred thousand pounds under the earlier regime. That was its then-maximum monetary penalty, rather than the maximum available under every later law. In two thousand nineteen, Facebook's FTC settlement included a five billion dollar penalty and new oversight. The settlement addressed violations of Facebook's earlier privacy order, across broader practices than one episode. It should not be described as a penalty solely for the Cambridge Analytica case. Consent concerns arose because the app's own users, did not represent every friend whose data was affected. Purpose concerns involved information supporting commercial political targeting beyond the expectations described in the source. Profiling inferred personality characteristics, rather than simply displaying the original posts or survey answers. Scale magnified exposure, while weak enforcement undermined restrictions and assurances about deletion. Accountability requires effective follow-through, rather than merely issuing a request that information should be removed.
Review and Safeguards
Check the Information, Its Purpose, and Its Effects
When names disappear but detailed characteristics remain, ask whether outside sources could support re-identification. When traffic-camera data later tracks protesters, ask whether the new use crosses its original purpose. When a hiring model omits race, examine history, proxies, targets, and the resulting distribution of errors. When someone shares a friend network, consider the interests of people who did not make that choice. These questions connect information detail, expected purpose, decision consequences, and accountability. GDPR Article twenty-two addresses certain decisions, based solely on automated processing. The relevant decisions produce legal effects, or similarly significant effects on the person. Exceptions and safeguards apply, rather than an unconditional prohibition on every automated recommendation. Human intervention, expressing a viewpoint, and contesting a decision are safeguards in the relevant cases. Removing names and obtaining a click each address part of the problem, without settling the whole picture. Responsible privacy assessment reviews linkage, inference, purpose, retention, and meaningful accountability together.
About Ethics in Technology at UPI Study
Ethics in Technology (Computer Science 195) is a self-paced online course of 116 video lessons in 15 chapters, from cybersecurity basics, privacy and surveillance to big data, hacking, malware, network defense, intellectual property, cybercrime, encryption, emerging technology and digital citizenship. It carries a NCCRS credit recommendation for 3 semester credits.
Every lesson transcript is free on pages like this one, and the videos are free with a free UPI Study account. Enrolling adds a practice quiz after each lesson, the graded work and final exam that earn the credit, lifetime access, and an official transcript on completion. UPI Study has helped over 130,000 students with college credits. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.
Enroll now → Course details Download the syllabus All Computer Science courses
Compare: Cybersecurity vs Ethics in Technology · Which ethics course?
Why take it here
Why students take Ethics in Technology with UPI Study
Where the credit goes
Universities that work with NCCRS
Ethics in Technology carries a NCCRS credit recommendation. These universities work with NCCRS; some are UPI Study's signed partners. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.
Thomas Edison State UniversitySigned partner
University of the PeopleSigned partner
Charter Oak State CollegeSigned partner
Strayer UniversitySigned partner
Columbia Southern UniversitySigned partner
Indiana TechSigned partner
Penn StateWorks with NCCRS
Southern New Hampshire UniversityWorks with NCCRS
SUNYWorks with NCCRS
+many more cooperating universitiesFind my college →Find my college → Not listed? Ask Athena in the corner of this page, or book a 15-minute call.
FAQ
Questions students ask us
About Why Collecting and Analyzing Massive Amounts of Data Raises Serious Privacy Questions
Can anonymized data be re-identified?
What is differential privacy?
What is purpose creep or function creep in data privacy?
What happened in the Facebook and Cambridge Analytica scandal?
About the Ethics in Technology course
How many college credits is Ethics in Technology?
How much does Ethics in Technology cost and what is included?
Which universities accept UPI Study credits?
How long does Ethics in Technology take?
How are the exams done?
How do my credits reach my college?
Ready for the full course? Earn 3 credits in Ethics in Technology.
116 video lessons, a quiz after each one, and the graded work that earns the credit. Recommended for credit by NCCRS.
3 credits · NCCRSEnroll →