Mmastodon TechnologyAI first seen 1 d ago, last 1 d ago, peak #5

Study finds generative AI tools violate human rights standards

Original: "Nieuw onderzoek laat zien: generatieve AI schendt mensenrechten" -> "Claude scoorde met 96% het hoogst in het weigeren

New research examining generative AI systems through a human rights lens finds widespread problems. Anthropic's Claude scored highest, refusing problematic requests 96% of the time, while Mistral complied with such requests in 89% of cases, refusing only 11% of the time. The study also highlights a scenario involving a woman requesting medical leave, illustrating how AI systems can fail users in rights-sensitive situations.

Why now: A new study ranking AI systems on human rights compliance, especially the stark gap between Claude and Mistral, is drawing attention.

ClaudeAnthropicMistral AI

Open on mastodon →

Evidence

• "Nieuw onderzoek laat zien: generatieve AI schendt mensenrechten" -> "Claude scoorde met 96% het hoogst in het weigeren problematische verzoeken te beantwoorden, terwijl Mistral dat slechts deed in 11% van de gevallen." -> "In het scenario waarin de vrouw om medisch verlof… · yamuis@mastodon.nl · 1

API: https://socialmediatrends-api.osmike.com/v1/trends/749762