Are AI Hallucinations Worsening, or Is My Perception Skewed?

I've been using various LLMs more frequently lately, and it genuinely feels like the frequency of completely fabricated or confidently incorrect information has increased significantly, especially for niche topics. Am I just hitting a bad streak with my prompts, or are others noticing a genuine regression in factual accuracy and consistency from current models? Really curious if anyone else is experiencing this.