Introduction to Fides Lang with Cillian Kieran

Summary

Kilian Kiernan introduces Fideslang, a high-level description language for privacy that simplifies complex privacy tasks by integrating with CI pipelines and runtime environments. Fideslang uses a standard ontology to define privacy characteristics of software systems and data, focusing on data categories, uses, subjects, and qualifiers. This language is designed to be human-readable, making it accessible for developers and non-engineers alike, ensuring privacy compliance and interoperability across systems.

Takeaways

• 📝 Fideslang is a high-level description language for privacy, designed to simplify and automate complex privacy tasks.
• 🌐 It integrates with CI pipelines and runtime environments, supporting privacy in software systems and data processing.
• 🔒 Privacy is complex due to varying global regulations and a lack of interoperability in data definitions across teams and companies.
• 👨‍🏫 Developers often require significant training to understand privacy concepts, highlighting the need for standardized privacy ontologies.
• 🗂️ Fideslang uses a taxonomy that includes four major privacy primitives: data categories, data uses, data subjects, and data qualifiers.
• 🔑 Data categories define the type of data processed, such as 'contact data' or 'email address'.
• 🎯 Data uses label the purpose of data usage, like 'advertising' or 'personalization'.
• 👤 Data subjects represent the type of user or subject whose data is being processed, with different rights and policies.
• 🔍 Data qualifiers denote the degree of identification in the data, such as 'aggregated', 'anonymized', or 'identifiable'.
• 📖 Fideslang is designed to be easy to read, understand, and write, making it accessible to developers and non-engineers alike.
• 🔗 It provides a clear understanding of privacy characteristics, behaviors, and risks associated with data processing in systems.
• 💡 Fideslang's hierarchical taxonomy and dot notation allow for precise and fine-grained privacy definitions, enhancing compliance and interoperability.

Q & A

What is Fideslang and how does it relate to privacy?

Fideslang is a high-level description language for privacy, supported by a set of tools and workflows. It integrates with CI pipelines and runtime environments to simplify and automate complex privacy tasks, helping teams manage privacy in compliance with global regulations.

Why is privacy a complex problem in software development?

Privacy is complex due to multiple global regulations with different definitions for how specific categories of data should be treated, compounded by a lack of interoperability as every team and company may define their own view of data types.

What does Fideslang aim to achieve by defining a standard ontology for privacy?

Fideslang aims to achieve a clear understanding of privacy characteristics, behaviors, and associated risks of systems' data processing by defining a standard ontology, making it easier for developers and non-engineers to understand and comply with privacy regulations.

What are the four major privacy primitives captured by Fideslang's taxonomy?

The four major privacy primitives are data categories, data uses, data subjects, and data qualifiers. These primitives help describe the type of data, the purpose of its use, the type of user or subject it pertains to, and the degree of identification it provides.

How does Fideslang's hierarchical taxonomy work?

Fideslang's taxonomy is hierarchical, allowing for both broad and fine-grained categorization of data. For example, 'contact data' is a broad category that includes all contact information, while 'email address' is a more specific sub-category within it.

What is the purpose of data categories in Fideslang?

Data categories in Fideslang represent the types of data a system processes, providing a clear 'what' aspect of the data being handled, which is crucial for understanding privacy implications.

Can you provide an example of how data uses are described in Fideslang?

Data uses in Fideslang are described using labels that indicate the purpose for which data is used, such as 'advertising' or 'personalization'. This provides a 'why' aspect, helping to clarify the intent behind data processing.

Why is it important to distinguish data subjects in privacy regulations?

Data subjects represent the type of user or subject whose data is being processed. Distinguishing them is important because rights or policies for data processing may vary by subject grouping, such as a customer versus a patient.

What do data qualifiers in Fideslang signify?

Data qualifiers in Fideslang denote the degree of identification for a given data, indicating how identifiable the individual is. Qualifiers include 'aggregated', 'anonymized', and 'identifiable', which help define the privacy level of the data.

How does Fideslang make it easier to comply with privacy regulations?

Fideslang provides an easy-to-read, understand, and write definition language for privacy that synthesizes major regulations. This ensures that rules and policies can be applied evenly across systems and provides an interoperable standard for privacy.

Where can one find more information about Fideslang and its technical documentation?

For more information about Fideslang, one can visit ethica.com or check the detailed technical documentation located on the GitHub repository for Fideslang.

Outlines

00:00

📜 Introduction to Fideslang for Privacy in Data Processing

Kylan Kieran introduces Fideslang, a high-level description language for privacy that integrates with CI pipelines and runtime environments to simplify privacy tasks. Fideslang addresses the complexity of global privacy regulations by providing a standard ontology for data privacy. It is designed to be human-readable and easy for developers to use, even those not familiar with privacy concepts. The language categorizes data into four privacy primitives: data categories, data uses, data subjects, and data qualifiers. These categories help in understanding the privacy characteristics, behaviors, and risks associated with data processing in software systems.

Keywords

💡Fideslang

Fideslang is a high-level description language for privacy that integrates with CI pipelines and runtime environments. It simplifies and automates complex privacy tasks. In the video, Fideslang is presented as a solution to the complex problem of privacy compliance due to multiple global regulations. It is designed to be easily readable and writable, making it accessible to developers and non-engineers alike.

💡Privacy

Privacy is a central theme in the video, referring to the protection of personal data and ensuring that it is handled according to legal and ethical standards. The video discusses the complexity of privacy due to varying regulations and the need for a standard ontology to describe privacy in data processing systems, which Fideslang aims to provide.

💡Data Categories

Data categories are types of data that a system processes. In the context of the video, they represent the 'what' aspect of data privacy—identifying the kind of data being handled. Fideslang uses data categories to classify information, which is crucial for understanding privacy characteristics and risks associated with data processing.

💡Data Uses

Data uses refer to the purposes for which data is utilized within a system. The video explains that these are labeled within Fideslang to describe why data is being used, such as for advertising or personalization. This concept helps in defining the context of data usage, which is essential for privacy compliance.

💡Data Subjects

Data subjects represent the individuals whose data is being processed. The video highlights that understanding who the data belongs to is critical because privacy rights and policies may vary by subject grouping. Fideslang accounts for this by categorizing data subjects to ensure appropriate data handling.

💡Data Qualifiers

Data qualifiers denote the degree of identification for a given data. The video explains that these qualifiers are crucial for determining how identifiable the individual is within the data. Examples include 'aggregated,' 'anonymized,' and 'identifiable,' which help in assessing the privacy risks associated with different types of data.

💡Taxonomy

A taxonomy in the video refers to a hierarchical classification system used by Fideslang to organize and categorize data and privacy concepts. It is used to maintain human readability and to structure the language in a way that is logical and easy to understand, which is essential for developers to implement privacy policies effectively.

💡Interoperability

Interoperability is mentioned in the context of the lack of a unified standard for data privacy across different teams and companies. The video suggests that Fideslang can help achieve interoperability by providing a standard ontology for describing privacy, allowing for consistent application of rules and policies across different systems.

💡CI Pipeline

CI (Continuous Integration) pipeline refers to the automated process of software development and testing. In the video, Fideslang is described as integrating directly with CI pipelines, which means it can be used to automate and simplify privacy tasks within the software development lifecycle.

💡Runtime Environment

Runtime environment is the context in which software operates and performs its functions. The video mentions that Fideslang integrates with runtime environments to simplify and automate privacy tasks, ensuring that privacy considerations are maintained even after software deployment.

💡Ethical AI

While not explicitly mentioned in the transcript, the concept of Ethical AI is implied through the discussion of privacy and data protection. Ethical AI involves ensuring that AI systems are designed and operate in a manner that respects privacy, fairness, and other ethical considerations. Fideslang contributes to Ethical AI by providing a framework for describing and managing privacy in AI systems.

Highlights

Fideslang is a high-level description language for privacy.

It integrates with CI pipelines and runtime environments to automate privacy tasks.

Privacy is complex due to varying global regulations and lack of interoperability.

Fideslang aims to define a standard ontology for privacy in data processing systems.

It is designed for describing privacy characteristics of software systems and data.

Fideslang is based on a taxonomy capturing four major privacy primitives: data categories, data uses, data subjects, and data qualifiers.

Data categories define the types of data a system processes.

Data uses label the purpose for which data is used, like advertising or personalization.

Data subjects represent the type of user or subject whose data is being processed.

Data qualifiers denote the degree of identification for given data, such as aggregated, anonymized, or identifiable.

Fideslang's taxonomy is hierarchical, allowing for both broad and fine-grained privacy descriptions.

It uses dot notation to indicate the relationship between data types and their categories.

Fideslang is easy to read, understand, and write, making it accessible to developers and non-engineers.

It synthesizes major regulations, reducing the need for extensive privacy training.

Fideslang ensures rules and policies can be applied evenly across systems.

It provides an interoperable standard for privacy in git repos, CI pipelines, and runtime environments.

For more information, visit ethica.com or the GitHub repository for detailed technical documentation.

Transcripts

[Music]

hi i'm kylian kieran and i want to talk

a little about fiji's lying today

feta is lying if you're not familiar is

a high level description language for

privacy supported by a set of tools and

workflows

it integrates directly with your ci

pipeline and runtime environment to

simplify and automate complex privacy

tasks

i'm going to share a little more about

how fideslang is used and the benefits

it can provide to any privacy conscious

team so privacy is a complex problem

because of multiple global regulations

with different definitions for how

specific categories of data should be

treated under certain circumstances

this problem is further compounded by a

lack of interoperability because every

team and company defines their own view

of what type of data they're dealing

with of course for developers who aren't

familiar with privacy even penetrating

these concepts means significant

training is required

so the solution to this is to define a

standard ontology for describing privacy

in data processing systems and that's

precisely what feed design is

so feed design is a high level

definition or description language

specifically designed for describing

privacy characteristics of software

systems their associated data sets and

external data sources and destinations

so in order to achieve this and maintain

human readability fides is based on a

taxonomy that today captures four major

privacy primitives which i'll explain in

a little more detail

the first of those is data categories

the categories are types of data that

our system is processing you can think

of this as the what right the what type

of data

the next is data uses or the taxonomy of

labels to describe the purpose for which

data is used in our system you can think

of this as the why

a good example of this might be

advertising or personalization

the third is data subjects so this is

the representation of the type of user

or sometimes called the subject in

privacy regulations so that's who whose

data we're dealing with

the user requires distinction because

rights or policies for how data is

processed may vary by the subject

grouping so an example might be a

customer of an e-commerce system versus

a patient in a clinical trials platform

the fourth and final grouping is data

qualifiers these are an attribute that

denotes the degree of identification for

a given data so that is the how

identifiable the individual is so

qualifier types include aggregated where

there's no individually identifiable

information anonymized which is data

that has been modified to remove

identifiable information or identifiable

which readily identifies the individual

by using these four resources together

as part of feed as lang we can build a

clear understanding of the privacy

characteristics behaviors and associated

risks of our systems data processing

this is the entire premise of feed-outs

and the tools that leverage the fetus

language so fides is very easy to read

understand and write

this is intentional as it should be easy

for any developer or even non-engineers

to pick up

the taxonomy is hierarchical so an

example could be contact data which

encompasses all of the contact

information a user might give to a

system or it might be more fine-grained

such as email address which is a

sub-category of contact information

so for example if we want to declare

that our system was processing data that

might identify a user we would simply

write

user

provided

identifiable data this dot notation

structure indicates that the data in

question was provided by the user and

identifies them directly

we could go a little further and state

that it's user provided identifiable

contact data so in this case we're

stating that it's part of information

related to the contact grouping

and finally we could be very precise and

fine-grained and we could state that

it's user provided identifiable contact

phone number and so in this case we're

clearly saying that it's the fundamental

related to the individual or user

as you can see fides provides an easy to

read and write definition language for

privacy that synthesizes major

regulations for you so you don't need to

in order to provide your work with a

readily understood

privacy compliance set of definitions

this ensures that rules and policies can

be applied evenly and ensures an

interoperable standard for privacy in

both your git repos your ci pipeline or

for evaluation of privacy requests in

your runtime if you'd like to learn more

please check out ethica.com

or the detailed technical documentation

located on the github feed outline

report