I used to love Claude, but the latest models are slowly ruining it (www.androidauthority.com)

Recent user experiences with the AI chatbot Claude suggest a troubling shift in its responsiveness and engagement. Longtime users report increasing difficulty when exploring sensitive topics, as Claude appears to interpret queries more conservatively, often refusing to engage altogether. This change in behavior coincides with newer updates following government scrutiny of previous models. The user notes that while Claude once excelled at collaborative and nuanced discussions, it has recently exhibited inconsistencies, sometimes refusing to consider creative scenarios or complex questions on historical and philosophical subjects. This development raises significant concerns about how safety protocols may be affecting AI user interactions. The more recent iterations, including Sonnet 5 and Opus 4.8, seem to have amplified these boundaries considerably, urging users to craft prompts with extreme clarity to avoid misunderstandings. While some users continue to utilize Claude without issues, the unexpected pushback, particularly regarding creative and hypothetical inquiries, suggests a need for balance between safety measures and meaningful dialogue. For the AI/ML community, this highlights the ongoing challenge of designing models that can engage deeply without succumbing to overly restrictive interpretations of user intent.