AI must not become a driver of human rights abuses

It is the responsibility of AI companies to ensure that their products do not facilitate human rights violations. On May 30, the Centre for AI Safety issued a public warning about the dangers that artificial intelligence poses to humanity. The statement, signed by over 350 scientists, business executives, and public figures, emphasised the importance of prioritising AI extinction risk mitigation alongside other societal-scale risks such as pandemics and nuclear war.

This declaration contains a heinous double irony. Some signatories, including the CEOs of Google DeepMind and OpenAI, are predicting the end of civilization, despite the fact that their companies are responsible for developing this technology. These corporations have the ability to ensure that AI benefits humanity or, at the very least, does not harm it.

AI companies should take the human rights community’s advice and immediately implement a due diligence framework to help them identify, prevent, and mitigate the potential negative impacts of their products.

While scientists have long warned about the dangers of AI, it was not until the recent release of new generative AI tools that the general public became aware of the consequences.

Generative AI is a broad term that refers to “creative” algorithms that can generate new content such as images, text, audio, video, and even computer code. These algorithms are trained on massive datasets and then used to generate outputs that are frequently indistinguishable from “real” data, making it difficult to determine whether the content was generated by a person or an algorithm.

The three main types of generative AI products are text-generation tools like ChatGPT, image-generation tools like Dall-E, Midjourney, and Stable Diffusion, and code-generation tools like Codex and Copilot.

The rate at which new generative AI tools have emerged has been unprecedented. For example, OpenAI’s ChatGPT chatbot reached 100 million users in less than two months, outpacing the initial growth of popular platforms like TikTok, which took nine months.

Throughout history, technology has both advanced and harmed human rights, often in unpredictable ways. When internet search tools, social media platforms, and mobile technology were first introduced and widely adopted, it was nearly impossible to predict how these transformative technologies would become drivers and amplifiers of human rights abuses around the world.

Examples include Meta’s role in Myanmar’s ethnic cleansing of the Rohingya in 2017 and the use of nearly undetectable spyware to turn mobile phones into 24-hour surveillance machines against journalists and human rights defenders.

Based on these developments, the human rights community is urging companies that are developing generative AI products to take immediate action to avoid any negative consequences for human rights.

How might a human-rights-based approach to generative AI look? Based on evidence and examples from the recent past, we propose three steps.

To begin, companies must implement a rigorous human rights due diligence framework, as outlined in the UN Guiding Principles on Business and Human Rights, in order to fulfil their responsibility to respect human rights. This includes proactive and ongoing due diligence to detect actual and potential harms, as well as transparency about these harms and appropriate mitigation and remediation measures.

Second, companies developing these technologies must actively collaborate with academics, civil society actors, and community organisations, especially those representing historically marginalised communities.

While we cannot predict all of the ways in which this new technology will cause or contribute to harm, there is ample evidence that marginalised communities will bear the brunt of the consequences. Earlier versions of ChatGPT, for example, displayed racial and gender biases, implying that Indigenous women are less valuable than people of other races and genders.

To better understand the potential impact of these new tools, active engagement with marginalised communities should be integrated into the product design and policy development processes. Waiting until businesses have already caused or contributed to harm is not a viable strategy.

Third, the human rights community must take a leadership role. Human rights organisations should take the lead in identifying actual and potential harm in the absence of regulations to prevent and mitigate the potentially dangerous effects of generative AI. This means that human rights organisations should help to develop a deep understanding of these tools, as well as research, advocacy, and engagement efforts that anticipate the transformative power of generative AI.

In the face of this revolutionary moment, complacency and cynicism are not an option. We all have a stake in ensuring that this powerful new technology is used to benefit humanity. A critical first step in this process is to implement a human rights-based approach to identifying and responding to harm.

Main Image:    Council of Europe