Global Leaders Sound Alarm: AI Threatens Human Rights Without Binding Safeguards

Author: Aamnah Fatima Khan

September 26, 2025

HRRC calls for the immediate implementation of safeguards to govern AI globally. Allowing AI development without oversight threatens to erode human dignity, freedom, and the very foundations of international human rights law.

The United Nations General Assembly (UNGA) in New York City, 2025 [Image source: DuckDuckGo Search]

A coalition of scientists, policymakers and Nobel laureates is warning that, without strict limits, AI could become a tool for violating human rights on a massive scale. Their joint appeal, unveiled at the United Nations General Assembly, demands enforceable international limits on AI applications by 2026.

The signatories argue that the unchecked spread of AI tools poses risks that go far beyond technological disruption. They warn of scenarios where governments or corporations could exploit AI for surveillance, manipulation or even lethal force, undermining democratic institutions and civil liberties. Concerns also extend to AI’s potential to worsen inequality through mass unemployment, destabilize societies with disinformation, and create irreversible security threats like engineered biological hazards.

What distinguishes this initiative from earlier appeals is its insistence on legal obligations rather than voluntary pledges. Previous industry promises, critics note, have proven to be unreliable, with compliance hovering just above half. Instead, the group is calling for clearly defined “red lines”, non-negotiable prohibitions on the most dangerous uses of AI–such as self-replicating systems–impersonation of individuals at scale, or automated control of nuclear arsenals.

Human rights advocates stress that this is not a rejection of AI’s potential. They acknowledge that, if properly governed, the technology could help address pressing challenges, from climate change to medical innovation. But they caution that without global safeguards, the same tools could be turned against citizens, eroding privacy, dignity and even life itself.

The coalition frames its deadline as urgent: with AI advancing rapidly, waiting past 2026 may leave little room for effective intervention. Their message is clear: humanity must act collectively to ensure technology serves people, not the other way around.

---

Glossary 

• Automated – Something that works by itself without human control

• AI – Artificial intelligence, machines or software that can think or act like humans

• Civil liberties – Basic personal freedoms that people are allowed to have in society

• Coalition – A group of people or organizations working together for a common goal

• Disinformation – False information spread to mislead people

• Enforceable – Something that can be carried out and made to happen by law or authority

• Engineered – Made or designed in a planned or controlled way

• Eroding – Slowly wearing away or weakening over time

• Exploit – To take unfair advantage of someone or something

• Frames – The way something is presented or explained

• Hazards – Things that are dangerous or cause harm

• Hovering – Staying in one position in the air or at a level without moving much

• Impersonation – Pretending to be someone else

• Insistence – Demanding something strongly and not giving up

• Invention – A new idea, tool, or process that has been created

• Irreversible – Something that cannot be changed back once it has happened

• Lethal – Something that can cause death

• Manipulation – Controlling or influencing people or things in a dishonest or unfair way

• Nuclear arsenals – Collections of nuclear weapons owned by countries

• Prohibition – A rule or law that says something is not allowed

• Rapidly – Happening very quickly

• Self-replicating – Able to make copies of itself without help

• Surveillance – Watching people or places carefully, often by cameras or monitoring systems

• Voluntary pledges – Promises people or organizations make by choice, not because they are forced

• Worsen – To make something worse

---

Sources

1. https://www.nbcnews.com/tech/tech-news/un-general-assembly-opens-plea-binding-ai-safeguards-red-lines-nobel-rcna231973 

2. https://www.webpronews.com/ai-pioneers-warn-un-of-catastrophic-risks-demand-global-red-lines-by-2026/ 

3. https://coingeek.com/over-200-leaders-experts-demand-global-red-lines-for-ai-use/ 

4. https://www.datamation.com/artificial-intelligence/global-ai-rules/