A coalition of human rights and technology groups on May 16, 2018 launched “The Toronto Declaration” which seeks to protect the rights to equality and non-discrimination in machine learning systems. The Declaration which was prepared by Amnesty International and AccessNow has also been endorsed by Human Rights Watch and Wikimedia Foundation.
The Declaration which was launched was at the annual meeting of digital and human rights groups: RightsCon conference in Toronto, Canada focuses on the rights to equality and non-discrimination considered critical principles underpinning all human rights.
It calls on States and private actors to promote the development and use of these technologies to help people more easily exercise and enjoy their human rights. This is against the backdrop of advance in capability and increase in use of machine learning systems and the critical need to examine the positive and negative implications of these technologies as they affect the enjoyment of human rights.
The Declaration aims to build on existing discussions, principles and papers exploring the harms arising from the use of this technology.
Relying on the framework of international human rights law, the Declaration holds that States have obligations to promote, protect and respect human rights and private sector (including companies) has a responsibility to respect human rights at all times.
It acknowledged the fact that “Human rights law is a universally ascribed system of values based on the rule of law which provides established means to ensure that rights, including the rights to equality and non-discrimination, are upheld. Its nature as a universally binding, actionable set of standards is particularly well-suited for borderless technologies such as machine learning. Human rights law provides both standards and mechanisms to hold the public and private sectors accountable where they fail to fulfil their respective obligations and responsibilities to protect and respect rights. It also requires that everyone must be able to obtain an effective remedy and redress where their rights have been denied or violated.”
It calls on all actors, public and private, to prevent and mitigate discrimination risks in the design, development and, application of machine learning technologies and ensure that effective remedies are put in place before deployment and throughout the lifecycle of these systems.
The Declaration underlines that inclusion, diversity, and equity are key components to ensuring that machine learning systems do not create or perpetuate discrimination, particularly against marginalised groups.
It lists three core steps to the process of human rights due diligence to include: identifying potential discriminatory outcomes; taking effective action to prevent and mitigate discrimination and document responses; and being transparent about efforts to identify, prevent, and mitigate against discrimination in machine learning.
The signatories to the Declaration called on the private and public sector to uphold their obligations and responsibilities under human rights laws and standards, in particular to avoid discrimination in the use of machine learning systems.
They also called on states and the private sector to work together and play an active and committed role in protecting individuals and groups against discrimination. They urged that when deploying machine learning systems, they must take meaningful measures to promote accountability and human rights including, but not limited to, equality and non-discrimination as per their obligations and responsibilities under international human rights law and standards.
They reiterated that technological advances must uphold human rights noting that: We are at a crossroads where those with the power must act now to protect human rights, including the rights to non-discrimination and equality – and help safeguard the human rights that we are all entitled to now, and for future generations.”
The Declaration is still open for sign-on and organisations, companies or governments wishing to endorse the Declaration are encouraged to contact firstname.lastname@example.org, email@example.com or firstname.lastname@example.org.