Open Ethics Transparency Protocol

Introduction

The Open Ethics Transparency Protocol (OETP or Protocol) describes the creation and exchange of voluntary ethics Disclosures for products that involve Automated Decision-Making (ADM), including AI-powered services, Robotic process automation (RPA) tools, and robots. It is brought as a solution to increase the transparency of how digital products with full or partial automation are built and deployed. This document provides details on how disclosures for data collection and data processing practice are formed, stored, validated, and exchanged in a standardized and open format.

What is Self-disclosure?

Digital products need to be transparent about how they collect, store, and process information for and about users. This becomes paramount when software product makes predictions or is integrated into the automated decision-making processes. Self-disclosure in the context of digital products refers to the practice of owners and deployers voluntarily sharing information about how their products are built and about how they are expected to behave while interacting with end-users or downstream software agents. Self-disclosure is essential as it helps build trust and tailor user experiences. Open Ethics designed the web form to help product owners disclose information about key pillars of their product including training data, algorithms, and the decision space. The tool generates a machine-readable JSON file, and the product owner receives the automatically generated HTML code with the visual representation of the disclosure (OEL), which can be pasted onto their website with zero effort.

Open Ethics Transparency Protocol

The protocol supports ethical disclosure of ADMs and provides facilities for:

  • Informed consumer choices : End-users able to make informed choices based on their own ethical preferences and the product’s disclosure.
  • Industrial scale monitoring : Discovery of best and worst practices within market verticals, technology stacks, and product value offerings.
  • Legally-agnostic guidelines : Suggestions for developers and product-owners, formulated in factual language, which are legally-agnostic and could be easily transformed into product requirements and safeguards.
  • Iterative improvement : Digital products, specifically, the ones powered by artificial intelligence could receive a nearly real-time feedback on how their performance and ethical posture could be improved to cover security, privacy, diversity, fairness, power-balance, non-discrimination, and other requirements.
  • Labeling and certification : Mapping to existing and future regulatory initiatives and standards.

Glossary book is only visible on a larger screens. If you’re on mobile, you can still explore it here

The Open Ethics Transparency Protocol (OETP) is an application-level protocol for publishing and accessing ethical Disclosures of Products and their Components. The Protocol is based on HTTP exchange of information about the ethical “postures”, provided in an open and standardized format. The scope of the Protocol covers Disclosures for systems such as Software as a Service (SaaS) Applications, Software Applications, Software Components, Application Programming Interfaces (API), Automated Decision-Making (ADM) systems, and systems using Artificial Intelligence (AI). OETP aims to bring more transparent, predictable, and safe environments for the end-users. The OETP Disclosure Format is an extensible JSON-based format.

The Internet Draft of the protocol document is available on GitHub. The URI scheme is available in this repository URI scheme.

Preview of the Protocol documentation

Get involved

Fields of the Transparency Disclosure in OETP: JSON schema

Schema JSON Schema for a Transparency Disclosure (https://openethics.ai/schema/oetp/oetp.disclosure.schema-v0.9.3.json) defines object using standard schema specification https://json-schema.org/draft/2020-12/schema
Name
Type
Description
Required
schema
object
JSON Schema for a Transparency Disclosure stored in the oetp.json of the Open Ethics Transparency Protocol v0.9.3 RFC. Will be deprecated after the update to v1.0.0 Beta
required
schema
object
The metadata properties about the schema, filled in automatically. This element will be deprecated in the next version of the schema because the disclosure manifest will have a direct reference to the Schema file.
required
name
string
The name of the schema
required
version
string
The version of the schema
required
integrity
string
The integrity hash is generated upon submission based on the serialized JSON of the submitted disclosure. The integrity hash is used to validate the disclosure and link the disclosure to the label via the HTML tag attribute.
required
snapshot
object
The snapshot of the disclosure data
required
product
object
The information about the product
required
url
string
The url of the product
required
description
string
The description of the product
optional
timestamp
number
The UTC timestamp of the disclosure
required
generator
object
The information about the schema generator, typically a disclosure provider
required
name
string
The name of the disclosure provider
required
alias
string
The alias of the disclosure provider
required
type
string
The type of the disclosure provider
required
website
string
The website of the disclosure provider
required
label
object
The object with the main elements of the disclosure
required
data
object
Please describe your data collection and data labeling practices, or your approach to heuristic in case if you are not using trained models for decision-making
required
type
string
Training Data Sources: Open Dataset, Proprietary Dataset, Limited Access Dataset, Rule-based heuristic
required
practice
string
The description of the training dataset handling and model training
optional
source
object
Please describe choices made for development and for use of existing code libraries
required
type
string
Code sources: Open Source Code, Proprietary Source Code
required
practice
string
The description of the practices of reusing the existing codebases or the information about the principles on which the proprietary code was built, without revealing of the sensitive information or the trade secrets
optional
decision
object
Please describe types of decisions/outputs the system generates and safety measures taken
required
type
string
Type of Decision Space: Restricted Decision Space, Unrestricted Decision Space
required
practice
string
Describe practices and approaches of how the outputs are handled, how the failure modes are defined and how the user will interact with the outputs of the digital product
optional
processing
object
Describe your data processing approaches, such as processing and retention of the Personal Data, Sources to external resources such as code, model description, and legal notices, provide clarity on output validation techniques including Human-in-the-Loop
optional
records
array
The list for the personal data records
optional
records item
object
A personal data record
optional
type
string
The type of the personal data record
required
source
string
The source of how the data record was obtained
required
retention
string
The personal data retention information
required
purpose
string
The purpose of the collection of the personal data records
required
consent
boolean
The information on whether the consent was obtained directly
required
identifying
boolean
The information whether the record is identifying (is PII)
required
source
array
The list with any additional sources of information
optional
source item
object
A unique source of the additional information
optional
type
string
The source type
required
url
string
The source url
required
comments
string
The comments about the source
optional
hitl
array
A list of the HITL approaches with their descriptions
optional
hitl item
object
HITL approaches description
optional
type
string
The type of the HITL
required
flow
string
The flow of how HITL is brought to the data processing pipeline
required
objective
string
The objective of having the HITL
required