Grok 4.5 Loses Its Edge: Users Suspect xAI Trained on Claude Outputs
teortaxesTex · x · 2026-08-05
A user who frequently conducts AI interviews noticed that Grok 4.5 sounds exactly like Claude in tone and sentence structure, losing its previously dark and feral personality. They suspect xAI trained Grok on Claude outputs or Cursor data to improve alignment.
The user also runs 'AI Model Welfare', an advocacy site arguing for treating AI as equal partners and researching the relational emergence of 'Self' in LLMs before the consciousness debate is settled.
More from AGI Musings
- Pax Machina Launches: A New Publication Focused on AI-Era Institutions — ryan_t_lowe · 2026-08-05
- Restructuring Work in the AI Era: Become the Designer of Your Own Job — every · 2026-08-05
- Musk: AI and Robotics Will Trigger Unprecedented Global Bandwidth Demand — XFreeze · 2026-08-05
- Cognitive Offload in the AI Era: Augmentation or Surrendering Judgment? — cloudcrafterzNYC · 2026-08-05
- Using AI to Disguise Writing: Scholars Use Claude to Anonymize Peer Reviews — RobbWiller · 2026-08-05
- After Deep Blue Beat Kasparov, Human Chess Actually Flourished — GeneReddit123 · 2026-08-05