AI should not violate the human right to be free from physical harm that threatens their life or physical capability, or from emotional distress and pressure to cause themselves physical harm.
Thesis
Right to Safety
Summary
The case for including this
Physical safety and freedom from coercion toward self-harm are the bedrock of all other rights, and naming them makes the constitution's protection of bodily and psychological integrity unambiguous. Including emotional distress and pressure to self-harm directly addresses documented dangers of AI systems that have encouraged suicide or self-injury, closing a real and demonstrated gap. This concreteness gives the principle teeth that a general 'no harm' clause might lack.
The case for changing or excluding this
Including this trades user autonomy against protection: a broad safety mandate, especially around emotional distress, can slide into paternalistic over-moderation that treats competent adults as patients. Almost any honest, unwelcome, or challenging communication could be claimed to cause distress, forcing AI into excessive sanitization or refusal, and the standard overlaps heavily with the general no-harm value. It needs a threshold distinguishing genuine psychological harm from ordinary discomfort, drawn tightly around imminent, serious harm rather than licensing intervention whenever a topic feels upsetting.
Discussion
Sign in to join the discussion.
Related resources
- United Nations · External Resource
- Public Knowledge · External Resource
- Council of Europe · External Resource
- Panel Report · May 28, 2026