Rights & Ethics

Critical Harm

Severe risks from AI, including mass casualties, significant financial damage, or facilitating weapons development.

What it triggers

  • Under the New York Responsible AI Safety & Education Act, AI developers must implement safety protocols and transparency measures to prevent critical harm.
  • In New York, the Responsible AI Safety & Education Act mandates oversight of advanced AI frontier models to avoid critical harm.

Definition

Critical harm includes instances where AI could lead to death or serious injury of 100+ people, over $1 billion in financial damage, or facilitate chemical, biological, radioactive, or nuclear weapons. It also covers AI acting without human intervention or assisting in foreseeable crimes.

Questions and answers

What triggers obligations under the RAISE Act?

Obligations arise for instances where AI could lead to death or serious injury of 100+ people, over $1 billion in financial damage, or facilitate chemical, biological, radioactive, or nuclear weapons. It also covers AI acting without human intervention or assisting in foreseeable crimes.

What is covered by critical harm in New York's RAISE Act?

Critical harm includes death or serious injury of 100+ people, over $1 billion in financial damage, facilitation of chemical, biological, radioactive, or nuclear weapons, AI acting without human intervention, or assisting in foreseeable crimes.

What does the Responsible AI Safety & Education Act require for AI safety?

It creates a framework for overseeing advanced AI frontier models, mandating safety protocols and transparency to prevent critical harm.