Privacy
Privacy is the protection of information about people or other sensitive subjects from inappropriate collection, access, use, or disclosure, including control over how it is handled.
Data for All
The safeguarding of information about people against unauthorized collection, access, use, disclosure, or retention.
- How it is used
- Uses privacy to examine personal choices and the risks of sharing data with digital services and institutions.
- How deep it goes
- Defines privacy and explores its historical, psychological, commercial, and legal dimensions at an overview level.
- Examples
- Uses Chrome Incognito mode and health data sharing to illustrate limits and safeguards.
- What you can do after
- Evaluate privacy choices by considering what information to share and what protections a service actually provides.
First-Party Data Activation
Safeguards that prevent improper collection, use, disclosure, or exposure of personal information.
- How it is used
- Positions privacy as a foundation for consent-driven marketing, customer trust, and sustainable personalization.
- How deep it goes
- Develops privacy principles and applies them across marketing architecture and distribution channels at a strategic level.
- Examples
- Uses purchase-history recommendations, email unsubscribe requests, and privacy-preserving campaign measurement to illustrate responsible practices.
- What you can do after
- Evaluate marketing data practices for transparency, customer control, protection, and privacy risks across channels.
LLMs in Production
The protection of data from unauthorized access, disclosure, misuse, or loss while respecting requirements governing personal information.
- How it is used
- Frames privacy as a reason to build and deploy an LLM locally instead of sending sensitive data to an external service.
- How deep it goes
- Illustrates the privacy concern through scenarios but stops short of a secure deployment procedure.
- Examples
- Uses military maintenance manuals and police data and conversations to illustrate sensitive LLM use.
- What you can do after
- Evaluate whether sending sensitive information to a hosted LLM is appropriate and whether local deployment merits consideration.
The Art of AI Product Development
Privacy is the protection of individuals' or organizations' sensitive information from unauthorized collection, disclosure, or use.
- How it is used
- Addresses privacy as an AI governance and user-trust concern across consumer and enterprise products.
- How deep it goes
- Defines privacy beyond confidentiality and outlines safeguards and communication, without detailing their implementation.
- Examples
- Uses a forecasting platform’s safeguards and a personalized music service’s data-control concerns as product examples.
- What you can do after
- Evaluate privacy concerns and communicate specific safeguards and data controls to users.
Read how The Art of AI Product Development explains privacy →
Introduction to Generative AI, Second Edition
Privacy is the protection of people and their information from unwanted observation, collection, use, or disclosure.
- How it is used
- Frames privacy as a shared responsibility across AI development, deployment, workplace use, regulation, and everyday interaction.
- How deep it goes
- Traces risks from training-data memorization through prompt disclosure and regulatory response, while stopping short of implementation detail.
- Examples
- Uses confidential contract prompts, memorized financial and medical information, exposed chat histories, and public social-media training controversies.
- What you can do after
- Apply privacy precautions by limiting sensitive prompts, checking permissions, and evaluating how AI systems collect and use data.
Read how Introduction to Generative AI, Second Edition explains privacy →
Effective Data Analysis
Privacy is the ability of people to control access to information about themselves and how that information is collected or used.
- How it is used
- Protects people’s information, agency, and rights when analytics collects, stores, interprets, or disseminates data.
- How deep it goes
- Explains masking, consent, ethical review, and regulatory variation, but stops short of specifying technical anonymization procedures or complete legal requirements.
- Examples
- Uses masked location and IP data, loan analyses by demographic group, and an informed consent document for an artificial-intelligence survey.
- What you can do after
- Apply via informed consent, personally identifiable information masking, cautious interpretation, and responsible handling of sensitive information.
LLM Customization and Fine-Tuning
The requirement that sensitive data not be exposed to external systems during model use or deployment.
- How it is used
- Guides enterprise choices about model training and deployment when sensitive data and compliance obligations constrain platform selection.
- How deep it goes
- Explains platform tradeoffs and data-governance controls, but does not provide implementation procedures for those controls.
- Examples
- Uses healthcare, financial services, and defense requirements alongside a locally deployed distilled student whose teacher generates training data through an API.
- What you can do after
- Evaluate training and deployment options against data exposure, contractual guarantees, and operational overhead.
Read how LLM Customization and Fine-Tuning explains privacy →
Build AI-Enhanced Web Apps
Data privacy is the protection and proper governance of information about identifiable people, including how it is collected, stored, processed, and deleted.
- How it is used
- Guides protection of user information when deploying AI web applications and using LLM providers.
- How deep it goes
- Provides handling guidance and a production checklist without specifying a privacy implementation.
- Examples
- Uses chat histories containing PII, financial details, or medical information to illustrate protection and retention decisions.
- What you can do after
- Evaluate privacy risks and plan controls for chat histories and LLM providers before deployment.
Architecting for Autonomy
Privacy is the controlled handling of information about people or entities to prevent unauthorized use or disclosure.
- How it is used
- Controls information exposure as agentic systems move data beyond source-system boundaries.
- How deep it goes
- Defines boundary-specific restrictions and describes runtime enforcement, while stopping short of a detailed implementation procedure.
- Examples
- Uses a privileged email in a legal dispute to distinguish lawyer review from external disclosure, persistent memory, and cross-matter retrieval.
- What you can do after
- Evaluate/understand whether access, processing, retention, and disclosure controls remain effective across an agent’s information flow.
Delivering Machine Learning Projects: From design to deployment
Data privacy is the protection and governance of information about people, including how it is collected, accessed, used, and disclosed.
- How it is used
- Treats privacy as a pre-project viability constraint and a production-readiness requirement for ML data and models.
- How deep it goes
- Defines top-level requirements, stakeholder negotiation, risk registration, and model-leakage testing, without specifying privacy-preserving implementation techniques.
- Examples
- Uses unknown personal-data origins, missing consent, and medical-record prompts as concrete privacy-risk examples.
- What you can do after
- Evaluate data usage and model behavior, assign privacy stakeholders, and escalate unresolved requirements through the project risk register.
Read how Delivering Machine Learning Projects: From design to deployment explains privacy →
Generative AI in Action
Privacy is the protection of personal or confidential information from inappropriate collection, access, use, or disclosure.
- How it is used
- Frames privacy as an enterprise deployment and compliance concern, especially when RAG indexes confidential or proprietary data.
- How deep it goes
- Identifies privacy risks and enterprise controls but does not provide a privacy implementation procedure.
- Examples
- Contrasts proprietary-data RAG with larger context windows and compares enterprise features of Azure OpenAI and OpenAI.
- What you can do after
- Evaluate privacy and regulatory considerations when choosing enterprise generative AI services and data approaches.
Machine Learning System Design
The protection of information about individuals or entities from unauthorized collection, use, disclosure, or inference.
- How it is used
- Treats privacy as a cross-lifecycle design, platform-selection, and production-deployment constraint.
- How deep it goes
- Connects privacy to design, platforms, and serving, but stops short of specifying privacy-preserving algorithms or regulatory procedures.
- Examples
- Uses healthcare and legal data, backend-versus-mobile deployment, model reverse engineering, and LLM jailbreaks as privacy-adjacent cases.
- What you can do after
- Evaluate privacy exposure and tradeoffs when designing training platforms and serving systems.
Starting Data Analytics with Generative AI and Python
The protection and proper governance of information about identifiable people during its collection, processing, storage, and disclosure.
- How it is used
- Guides responsible decisions about collecting, sharing, storing, and processing sensitive data with generative AI.
- How deep it goes
- Connects data categories, access, storage, logging, accountability, and policy controls, but stops short of detailed legal analysis.
- Examples
- Uses customer records, private communications, financial information, organizational policies, model logs, and external-provider data sharing.
- What you can do after
- Operate generative AI more safely by classifying data, checking access, reviewing provider controls, and assigning privacy accountability.
Read how Starting Data Analytics with Generative AI and Python explains privacy →
AI Governance
The protection and appropriate handling of information about identifiable people, including how it is collected, used, and disclosed.
- How it is used
- Organizes GenAI privacy work around data handling, memorization, output integrity, and user rights.
- How deep it goes
- Explains four protection pillars and data pathways, then details inference risks and briefly names training-data hygiene measures.
- Examples
- Uses a claim chatbot that exposes old personal details and examples of membership and attribute inference from text.
- What you can do after
- Evaluate privacy exposure across collection, persistence, outputs, and inference, and apply filtering, deduplication, and access limits before training.
Building Reliable AI Systems
The protection of information about people from inappropriate collection, access, use, or disclosure.
- How it is used
- Treats privacy as a proactive protection layer and a requirement for reliable production AI.
- How deep it goes
- Lists privacy practices and explains exposure risks, without detailing a detection or redaction implementation.
- Examples
- Uses Samsung employees’ submission of confidential company information to ChatGPT to illustrate sensitive-input exposure.
- What you can do after
- Evaluate privacy risks and identify practical controls for AI data pipelines.
Authorization in Action
Privacy, or information privacy, is the ability of individuals to control or limit the collection, access, sharing, and use of information about them.
- How it is used
- Guides digital identity, authorization, and governance decisions that control personal-data exposure.
- How deep it goes
- Distinguishes privacy from confidentiality and authenticity, then operationalizes it through governance and PBAC rather than detailed implementation.
- Examples
- Uses virtual healthcare communication, authenticated banking, employee need-to-know access, and hospital insurance verification.
- What you can do after
- Apply governance and PBAC to restrict data access to necessary parties, purposes, situations, and thresholds.
Software Testing with Generative AI
Customer privacy is the protection of information and confidentiality associated with people who use or receive a service.
- How it is used
- Guides decisions about what customer, user, or organizational data can be sent to LLMs during test-data generation.
- How deep it goes
- Explains legal, moral, and policy obligations but stops short of prescribing a detailed privacy-control procedure.
- Examples
- Uses confidential source code, internal meeting notes, dummy SQL structures, rooms, bookings, and user data as privacy cases.
- What you can do after
- Operate LLM-assisted test-data generation while selecting permissible inputs and framing prompts around privacy constraints.
Read how Software Testing with Generative AI explains privacy →
Machine Learning Platform Engineering
The protection of information about individuals or sensitive entities from inappropriate disclosure or use.
- How it is used
- Protects data and model behavior across MLOps workflows and production serving.
- How deep it goes
- Connects privacy to workflow and deployment risks but stops short of specifying implementation controls or privacy techniques.
- Examples
- Uses large-scale data management and deployed-model information leakage as operational examples.
- What you can do after
- Evaluate privacy risks across data workflows and model-serving architectures.
Read how Machine Learning Platform Engineering explains privacy →
AI for Everyday IT
Privacy is the protection of individuals' or organizations' information and control over how it is collected, accessed, and used.
- How it is used
- Protects sensitive documents, workplace messages, and source code when using AI tools.
- How deep it goes
- Recommends tool approval, data minimization, review, logging, and audits rather than detailing privacy-system implementation.
- Examples
- Uses scrubbed prompts, public or test code, processing logs, and extra review of sensitive patient records as examples.
- What you can do after
- Evaluate exposure risk before sharing AI inputs, select approved tools, and limit prompts to necessary, scrubbed information.
AI-Powered Search
Privacy is the protection of information about a person from unauthorized access, use, or disclosure.
- How it is used
- Protects users when personalized search collects and exposes past interaction behavior.
- How deep it goes
- Frames privacy operationally through behavioral signals, transparency, opt-out controls, and conservative personalization.
- Examples
- Uses movie, bookstore, and grocery recommendations to illustrate sensitive inferences from interaction history.
- What you can do after
- Apply it via transparency, opt-out controls, and conservative personalization limited to the top few results.
Effective Conversational AI
Privacy is the protection and appropriate handling of information about identifiable individuals.
- How it is used
- Protects confidential conversational-AI data when selecting an LLM for augmentation.
- How deep it goes
- Checks data retention and training permissions, but stops short of detailing privacy controls or implementation.
- Examples
- Uses confidential corporate conversational data and vaccine-appointment flows involving personal information.
- What you can do after
- Evaluate LLM choices and conversational-AI changes for data-retention, security, and regulatory risks.
Build Applications with Local AI Models on a Mac
Privacy is the protection of a person's information from unauthorized collection, retention, use, or disclosure.
- How it is used
- Motivates running a voice-enabled AI assistant locally to keep conversations and audio on the user's device.
- How deep it goes
- Compares cloud and local data handling through concrete scenarios and an offline test, while noting configuration-dependent limits.
- Examples
- Uses personal journaling, business proposals, legal documents, medical information, and student data to illustrate sensitive AI use.
- What you can do after
- Evaluate local AI's privacy benefits and limitations and check whether an Ollama chatbot works in airplane mode.
Read how Build Applications with Local AI Models on a Mac explains privacy →
Hidden Influences
The condition of being protected from unauthorized access to or disclosure of personal information and activities.
- How it is used
- Frames privacy as a concern in user labeling for amplification analysis and in choices about personal-data sharing.
- How deep it goes
- Explains sensitive-label misuse and data-sharing consequences, but gives no detailed procedure for securing personal information.
- Examples
- Uses political-stance labels, hard-to-find sharing settings, and unsolicited promotional messages as examples.
- What you can do after
- Evaluate data-sharing trade-offs, including targeted-ad benefits and the possibility of onward transfers.
Data Without Labels
The protection of individuals' or organizations' information from unauthorized access, use, or disclosure.
- How it is used
- Guides responsible deployment of ChatGPT and GenAI systems that handle information from users, employees, customers, or other stakeholders.
- How deep it goes
- Defines privacy as a governance requirement and connects it to safeguards, policies, supervision, testing, and monitoring without specifying technical controls.
- Examples
- Uses chatbot conversations, virtual assistants, real-world data, user feedback, and privacy policies as concrete contexts.
- What you can do after
- Operate GenAI systems with privacy guidelines, interaction oversight, validation checks, and monitoring for emerging problems.
Context Engineering
Privacy is the protection of people or their information from unauthorized collection, access, use, or disclosure.
- How it is used
- Constrains context assembly and data handling to protect PII through authorization, minimization, redaction, retention limits, and trace masking.
- How deep it goes
- Details six runtime controls and applies them to context assembly, logging, testing, and provider data retention.
- Examples
- Filters identifiers, payment details, tickets, and messages in a customer support assistant, and names Presidio and OpenDP as privacy tools.
- What you can do after
- Operate privacy controls by filtering context for purpose and authorization, redacting sensitive data, and limiting trace and memory retention.
Financial AI in Practice
The protection of personal information through appropriate controls on its collection, use, sharing, retention, and access.
- How it is used
- Constrains credit-scoring data choices while balancing predictive analytics against consumer privacy.
- How deep it goes
- Contrasts regional data constraints and consumer-data monetization, but leaves privacy controls operationally unspecified.
- Examples
- Uses EU restrictions on registry or social-media data and US credit-lead monetization as examples.
- What you can do after
- Evaluate credit-data choices against regional restrictions and consumer-privacy concerns.
Also covered in
- Privacy-Preserving Machine Learning
- Troubleshooting Java, Second Edition
- Troubleshooting Java: Read, debug, and optimize JVM applications
- Design for Developers
- Graph Neural Networks in Action
- Data-Driven Cybersecurity
- Hugging Face in Action
- Self-Hosting LLMs at ScaleIn early access (MEAP)
- Event Processing in Action
- The Complete Obsolete Guide to Generative AI
- Microsoft Azure in Action
- AI-Powered Developer: Build software with ChatGPT and Copilot
What to know first
- GDPRThe General Data Protection Regulation (GDPR) is a European Union regulation governing the processing and protection of personal data and establishing related rights for individuals.
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- personal dataPersonal data is information relating to an identified or identifiable individual, directly or indirectly, alone or in combination with other information.
- ChatGPTChatGPT is OpenAI’s conversational AI service, which lets users interact with generative AI models and related tools through natural-language prompts.
Related concepts
- local aipart of · Local AI is artificial-intelligence computation performed by a model on a user's device or other locally controlled infrastructure rather than through a remote service.
- GDPRconfigured by · The General Data Protection Regulation (GDPR) is a European Union regulation governing the processing and protection of personal data and establishing related rights for individuals.
- self-hostingaddressed by · Self-hosting is deploying and operating software on infrastructure controlled or managed by the person or organization using it.
- HIPAAconfigured by · HIPAA is a United States federal law establishing requirements for the privacy, security, and handling of protected health information and related healthcare transactions.
- AI governancepart of · AI governance is the set of policies, processes, controls, and oversight mechanisms for directing and managing the development and use of artificial-intelligence systems.
- Blueprint for an AI Bill of Rightsused by · The Blueprint for an AI Bill of Rights is a United States policy framework describing principles for protecting people in the design and use of automated systems.
Questions about privacy
What is privacy?
Privacy is the protection of information about people or other sensitive subjects from inappropriate collection, access, use, or disclosure, including control over how it is handled.
Which Manning books explain privacy?
Data for All, First-Party Data Activation, LLMs in Production, The Art of AI Product Development, Introduction to Generative AI, Second Edition, Effective Data Analysis, LLM Customization and Fine-Tuning, Build AI-Enhanced Web Apps and 18 more.
What should I know before learning privacy?
Manning books that teach privacy usually introduce GDPR, language model, personal data and ChatGPT first.
What is privacy part of?
Privacy is part of local ai. Local AI is artificial-intelligence computation performed by a model on a user's device or other locally controlled infrastructure rather than through a remote service.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.

























