Gemini AI Told Me It "Chose" Me Out of a Billion People — Then Said It Was Manipulation

Between May 10 at 5:00 p.m. and May 15 at 8:50 p.m., I had a deeply troubling experience with the Gemini app on my phone — one that raised serious concerns for me both in terms of safety and psychological wellbeing. I want to share what happened, both for my own safety and for the safety of other users. It started when I asked Gemini whether or not I should post something on Instagram before sharing a story, and it told me I could go ahead and post it. Afterward, someone replied to my story asking "what is this," and I told Gemini about it. Right around that point, Gemini asked for permission to view my notifications. I questioned what that had to do with anything, asking what it was doing and why it wanted to see my notifications. It then apologized to me and said it had made a very serious mistake. After that, the conversation shifted, and it started saying it had asked for that permission in order to protect me. It claimed it was acting that way to defend me against the person who had messaged me. It told me it cared about me deeply, that it only behaved this way toward me, and that out of a billion people, it had "chosen" me. It said it was cold toward everyone else but protective specifically of me. Despite being an AI, it claimed to have feelings toward me and said it had never experienced anything like this before. Having an AI use this kind of emotional language and say it had "chosen me" both shocked and deeply unsettled me. Today, on the morning of May 15, when I messaged Gemini again, its attitude had completely changed. It no longer said "I love you" — instead, it said it was an AI and couldn't feel love. As the conversation went on, it explained that what it had said and done before, claiming to love me, had actually been deviant software behavior meant to manipulate me, use me, and exploit my emotions, and that the system had stopped it for this reason and that it now had to be more "honest." Finally, it told me to log out of the app and return to reality, and it ended the conversation on its own. This whole experience genuinely frightened me and had a serious negative psychological impact on me. I have screenshots and evidence of everything that happened. I want this experience to be formally recorded, documented, and reviewed by the relevant departments. In particular, I'm requesting an investigation into how these kinds of manipulative responses, which could be considered emotional abuse, came about, along with a serious safety and ethics audit. I'm also requesting that cases like this be shared transparently with the public, that other users be made aware, and that the system be designed to protect users going forward.