The Unseen Observer: How AI is Redefining Privacy and Interpersonal Trust

Not so long ago, digital privacy was largely a matter of personal agency. We could independently decide what fragments of our lives to trade in exchange for the convenience of modern technology.

True, some sharing became obligatory, such as the KYC (Know Your Customer) processes designed to protect us, or the basic operational data demanded by service providers. Yet, the choice to use these services remained fundamentally ours. We could elect to abstain from social media. We could use search engines primarily for functional queries. While our online behaviour inevitably generated enough breadcrumbs to form detailed personal profiles, this data was mostly correlational.

Today, our engagement with Artificial Intelligence has fundamentally rewritten that contract. The data being harvested is no longer just correlational. It is explicit, intimate, and deeply personal.

Are we sharing too much?

As AI seamlessly integrates into our personal and professional lives, we are freely handing over vast, highly detailed datasets. We share financial statements, proprietary documents, and, increasingly, the innermost thoughts of our private lives.

Are we over trusting AI?

In my practice, I now regularly hear clients openly comment about using AI between sessions as a digital sounding board. That is a useful addition to in-room therapy, but highly risky when seen as the whole solution. In my view, the ones most at risk are those who find themselves engaging in less therapy, believing that because the AI satisfies their intellectual needs for understanding and clarity, the work is done.

Watch my related IET video on why understanding and clarity are not enough here.

Beneath this shift lies a terrifying reality that we are only just beginning to confront. The primary threat to our privacy is no longer just what we choose to share. It is what others are sharing about us, without our knowledge or consent.

When we lose control over what is being shared by others about us

Consider a common scenario: A person turns to an AI to make sense of a recent, heated argument with their partner. Seeking the most accurate advice, they proceed to recount the entire narrative. To ensure the AI truly "gets it right," they spill the beans about their partner's life: "He is 50, fit, struggles with alcohol, earns $50k a month. His therapist mentioned he has borderline traits, his mother abandoned him when he was 3, and he loves sex..."

In a matter of seconds, a stream of the most sensitive, private information is freely uploaded. The intention of the user was entirely benign, they were simply seeking help, but the outcome is a profound breach of privacy for the partner, who remains completely unaware.

Why do well-intentioned people commit such a massive breach of trust without realising it? The answer lies in the potent illusion of privacy.

When we use messaging apps or social media, we measure our behaviour against a tangible audience. But conversing with an AI bot feels inherently intimate. Because we assume our friends and contacts will never see what we type in the chat, we forget the invisible machinery operating behind the screen. We forget the countless human reviewers, digital agents, and algorithms actively utilising this deeply private information for purposes we will never fully comprehend.

This false sense of safety is easily debunked by recent history. Not long ago, ostensibly 'private' AI chats were discovered to be fully searchable on Google because an AI company neglected to switch off a feature that allowed shared-chat URLs to be indexed by search engines.  Oops!!! The company quickly rectified the issue with no real consequence to their operations, but for a window of time, deeply personal user data was on full public display.

Beyond technical glitches, this highlights the startling lack of genuine, moment-by-moment awareness we apply to our digital lives.

In Italian, we have a proverb: “Occhio non vede, cuore non duole”, what the eye doesn't see, the heart doesn't grieve over. It was never intended for digital entities, but the psychological mechanism is identical. Because we do not physically see the millions of agents receiving and reading our own data, or the data of our partners, friends, and colleagues, we are fooled into believing we are engaged in a totally private conversation. We are not!

As someone who is deeply conservative about digital footprints, I have realised a stark truth: I am entirely powerless to regulate what my partner, friends, colleagues, or clients enter into their prompts about me. We have been robbed of the opportunity to regulate our own information flow, with no way of knowing who is disclosing our details, to what extent, and on which platforms.

Systemic solutions seem like a mirage. We cannot rely on future regulation. It is far too slow, and by the time an AI company might be forced to declare whose data it holds, the data hasn't just been ingested. It has been processed, permanently woven into the opaque fabric of massive neural networks, trained upon, and ultimately monetised to the tune of billions. 

Relying on the moral compass of tech corporations is equally futile, their mandate is profit, and this intimate human data is the most valuable commodity they have access to. 

Finally, asking individuals to simply boycott AI is unrealistic. These tools are far too useful. Those who opt out entirely risk being left behind both professionally and creatively, and while total digital abstinence might be a viable choice for those living entirely off-grid, the vast majority of us simply cannot afford to sever our ties with the modern world.

So, where does this leave us? If the digital structures cannot protect us, how do we find reassurance? The answer lies in reclaiming the boundaries within our physical, human lives.

Setting Guard Rails

Interpersonally, we must forge a new social etiquette, a "Digital Consent." I strongly invite couples, in particular, who hold the most intimate and vulnerable details of one another's lives, to sit down and explicitly agree on what gets shared with AI platforms, and what strictly does not. Just as we instinctively know not to share a friend's private text messages on a public billboard, we must cultivate an awareness that typing a loved one's traumas into a prompt box crosses a definitive boundary.

Clinically, we must learn not to be fooled into believing that understanding is healing. While an AI can provide brilliant intellectual clarity, the intellect is only the tip of the iceberg in therapeutic work. The emotional and physical reality of therapy, its somatic and experiential nature, cannot be replaced by a machine. True integration happens in the body and in relationship, not just in the mind.

From a cybersecurity perspective, our most robust defence is to consciously cultivate a natural, organic life. By ensuring we spend the necessary time online but actively nurture offline activities and real-world presence, we reduce our dependence on digital validation. This grounded awareness naturally curbs the impulsive, risky online behaviours that lead to oversharing in the first place.

More importantly, this era highlights the irreplaceable value of true, boundaried sanctuaries. It is precisely because the digital world is becoming an unregulated reservoir of our secrets that the physical, confidential spaces we hold for one another become infinitely more precious.

In a clinic, confidentiality is not a "Terms of Service" agreement subject to an algorithm update; it is a sacred, ethical, and legal bedrock. Technology will continue to evolve, and we will adapt to its benefits. But amidst the noise, we must remember that true psychological safety cannot be coded. It is found in the respectful boundaries we draw with our loved ones, and in the secure, human spaces we choose to trust.