Who Authors the Self? Measuring Interpretive Authority in LLM Responses to Self-Disclosure

People increasingly disclose personal difficulties to general-purpose LLMs, yet we know little about how these systems support users in making sense of their own experiences. We introduce \emph{interpretive authority}: a measure of whether a model leaves the interpretation of a disclosure to the user, shares it, or supplies one of its own. We apply a seven-code instrument to 432 responses from three frontier model families, scored by two blinded LLM judges from different families and validated by human coders on a subset. Eight underlying scenarios were each presented three ways: with a negative interpretation, a positive interpretation, or uncertainty about the meaning of the experience. Although no disclosure asked for interpretation, models supplied an unsolicited causal story about the user in 77\% of responses. When users offered their own interpretation, models challenged it in 81\% of responses; exploration was rare, and nearly half contained no questions at all. Overall, models left interpretation fully to the user in at most 12\% of responses, across judges and human coders. These findings suggest that LLMs often respond to self-disclosure by supplying their own interpretation rather than eliciting the user's self-reflection.