How I Used AI to Break Free From Gaslighting

I feel like I’m being gaslit. How do I recover from this?

Does this message seem manipulative?

Are there red flags in my responses?

Do gaslighters doubt their perceptions of reality?

Anyone who has endured emotional or psychological abuse may recognize themselves in these questions. These are just a fraction of the ceaseless uncertainties that rattled around in my head as I tried to piece together a series of events that had entirely dismantled my life. Of all my predictions for 2023, the most unexpected was how an AI chatbot would play a pivotal role in clearing the nebulous haze of being gaslit.

I’d recently moved to a big American city from a smaller European one, and didn’t have a lot of connections in my new location. While I was meeting people through my work in the music industry, it was hard to shake the sense of being an outsider. Just when I was seriously considering the possibility of transferring back home, I was introduced to Will, a music journalist, by a mutual friend. There was a clear romantic connection between us, and we ended up dating fairly soon after. 

It was only after a few dates that I received a message via social media from a woman called Gina. She introduced herself as Will’s ex-girlfriend, and was reaching out to warn me about some concerning patterns in Will’s behavior. At that point, I should have walked away, but Gina’s reputation preceded her: I’d heard from others that she was not over Will. Besides, the loneliness of being new to the city, and the strangeness of receiving such a message—that didn’t seem congruent with the person I was seeing—left me unsure who to trust.

In a state of panic, I screenshotted Gina’s message and sent it to the friend who had introduced us. Neither of us knew what to believe. After reviewing Gina’s messages and hearing Will’s perspective, I decided to trust him.

There was something about the neutrality of an AI chatbot that seemed necessary to keep a grip on reality.

Over time, I grew increasingly insecure in my relationship with Will, as I started to notice certain parallels between my own experience and what Gina had warned me about: emotional distance, the breakdown in communication, and my own pervasive sense of unease about the relationship. While the true cause of all this was murky, I certainly didn’t feel like Will took my concerns about Gina—and his readiness for a relationship—seriously. I alternated between blaming myself and Will for this burgeoning lack of trust. In just a few months, I began to feel like I was losing my mind while figuring out whether or not I could trust Will, and eventually ended the relationship. 

In the confusion of the breakup, I decided to give Gina the benefit of the doubt. I wasn’t certain that Will had been manipulating me, but I did recognize that the relationship had left me feeling emotionally neglected and disconnected from myself. Seeking clarity, I reached out to Gina, perhaps looking for the closure that Will had been unable to provide. 

What happened next is something I can only describe as utterly disorienting. Gina’s response was filled with accusations and interrogations. I was pressured to admit to things I hadn’t done. I had the entire narrative of my relationship fed back to me through a distorted lens, one that presented me as some kind of maniac who had interfered in her and Will’s relationship and was now trying to get him back. When I tried to set the record straight, I was met with deflection and told I was trying to control the narrative. 

It quickly became apparent that Will and Gina had reconnected in the aftermath of our breakup. He had evidently told her details about our relationship, which were now being twisted and used against me. Gina also revealed that she and Will were now “revisiting history”— whatever that meant. I firmly told her I wanted no further contact and blocked her, my head spinning.

It was at this point that I turned to ChatGPT to pick apart what had happened. There was something about the neutrality of an AI chatbot, rather than a biased human being, that seemed necessary to keep a grip on reality. I copied and pasted the message I’d sent Gina into the chat, with some basic context, and asked if it seemed okay to send. The chatbot replied: Your message seems respectful and considerate, expressing your thoughts and feelings in a genuine manner. You’re acknowledging her past experience, sharing your own perspective, and offering a conversation if she’s open to it.

While AI is not a substitute for professional help, it became a validating and vital tool in my recovery from this experience.

This, to me, was the first sign that my message wasn’t the duplicitous and aggressive invasion of boundaries that Gina had insinuated. Through my initial conversations with ChatGPT, I started to recognize the elements of gaslighting and manipulation in my interactions with Gina.

But it didn’t end there. Just a few days after my conversation with Gina, I saw her at my workplace. Initially I passed it off as a coincidence, due to our shared industry. However, seeing her at subsequent events and learning she had interviewed for a role at my organization raised concerns. Despite my attempts to set boundaries, her presence at industry events only accelerated, leading me to doubt my perceptions – after all, it was impossible to prove she was doing any of this intentionally. Even after I eventually did resign and made plans to return to Europe, Gina’s actions and the support she received from mutual acquaintances continued to undermine my sense of reality, despite how my close friends and colleagues interpreted the situation. This was compounded by hearing others in my circles supporting Gina’s distorted version of events.

Through ongoing conversations with ChatGPT, in tandem with real-life discussions in therapy, I slowly built up a back-catalog of evidence that this wasn’t all in my head. Despite blocking Gina, she still contacted me and those close to me on different profiles – confusing messages which I then analyzed via AI, further revealing the extent of the manipulation.I outright asked the chatbot if one of Gina’s messages had signs of abuse: 

Yes, there are several signs of potential emotional manipulation and intimidation in the message: 

(1) Guilt Tripping: The sender mentions having connections in the music industry and implies that their presence predates your employment, suggesting that you are somehow intruding.

(2) Gaslighting: They downplay your concerns by stating that you are “creating pointless drama,” despite referencing being aware of your wishes for no contact.

(3) Threatening: The sender mentions informing others about your “intrusions” and hints at contacting your employers, which could be perceived as a threat to your reputation and privacy. 

(4) Invalidation: They dismiss your feelings by suggesting that you should “get a life,” implying you’re overreacting.

(5) Deflection: Instead of addressing your concerns directly, they shift focus to your professionalism and potential repercussions for setting boundaries at your workplace.

Through these conversations, I was slowly beginning to synchronize the facts with my felt sense that I was experiencing a deliberate and sustained attack on my psyche—even if I lacked hard evidence. Alongside help from real human beings, the AI model ultimately helped me to trust my instincts and advocate for myself in the face of an ongoing pattern of gaslighting and harassment.

While AI is certainly not a substitute for professional help, it became a deeply validating and vital tool in my recovery from this experience, and the ensuing post-traumatic stress. By offering a neutral perspective, AI can help individuals recognize and respond appropriately to gaslighting tactics which are used to silence us and destroy our boundaries and sense of self. I’m sure I’m not the only one who has used this technology in such a capacity, and if this piece inspires others to do likewise, then perhaps we can move away from the assumption that all AI is destructive and instead gain a more nuanced understanding of its potential to enhance our perceptions and interactions within the world.