Popular search engine providers like Google and Microsoft have recently incorporated chat-based generative AI into their web browsers. Knowing that these products often provide inaccurate information, integrating disclaimers into the user interface for potential errors has become a common practice. However, their layout makes overlooking them likely. Little is known about the impact of these disclaimers relative to other message and content attributes. This study employs a paired-conjoint design to compare 108 variations of simulated interactions with a chat-based AI responding to hypothetical political search queries. Each stimulus varies along these four dimensions and their levels; Sources (high credibility vs low credibility vs high and low credibility vs absent), Information topic (security and defence of EU vs European migration and asylum politics vs control), Information veracity (true vs partially false vs false), and Verification disclaimer (in text vs verification tag vs absent). The study aims to determine the absolute (RQ1), relative (RQ2), and level-specific relative effects (RQ2a) of these attributes, with a particular focus on how the impact of verification disclaimers differs compared to these other attributes (RQ3).