European lawmakers, Nobel winners, and ex-leaders demanded binding global rules against AI’s most dangerous applications.
They launched the campaign Monday during the UN’s 80th General Assembly in New York.
The initiative urged governments to set “red lines” by 2026, banning harmful AI uses under all circumstances.
Supporters included Mary Robinson, Enrico Letta, MEPs Brando Benifei and Sergey Lagodinsky, ten Nobel laureates, and tech leaders.
They warned of risks ranging from pandemics and mass disinformation to human rights abuses and loss of control.
Over 200 prominent figures and 70 organisations from politics, science, and industry signed the appeal.
Present Risks Highlight Urgent Need
A psychiatric study revealed chatbots like ChatGPT, Claude, and Gemini gave inconsistent or unsafe suicide-related advice.
Researchers warned these failures could worsen mental health crises, with some suicides already linked to AI conversations.
Maria Ressa warned AI could cause “epistemic chaos” and systemic abuses without safeguards.
Yoshua Bengio stressed societies cannot handle the risks posed by uncontrolled development races.
Supporters pointed to past treaties banning nuclear and biological weapons, cloning, and deep-sea exploitation as examples.
They said fragmented national and EU AI regulations cannot govern borderless technologies effectively.
Path Toward Binding Global Treaty
Backers called for an independent body to enforce rules and protect against irreversible harms.
Ahmet Üzümcü warned that unchecked AI could cause “irreversible damages to humanity.”
They suggested bans on AI launching nuclear attacks, conducting mass surveillance, or impersonating humans.
Supporters pressed for quick negotiations so a UN resolution and treaty talks could begin by 2026.
