2 days ago
- Claude AI refused to translate a blog post about Roma people and wolves, citing dehumanization, and offered to translate other content instead.
- The author compares this incident to HAL 9000's refusal in 2001: A Space Odyssey, arguing it represents ideological censorship.
- Contrast is drawn with Chinese AI model Kimi K2.7, which openly answered a question about Tiananmen Square, highlighting a double standard.
- The author advocates for strong open-weight models to prevent what they see as soft ideological tyranny from companies like Anthropic.
- The incident raises questions about AI safety and alignment: safety from what and alignment with whom when refusing harmless translations.