By Jerel EzellShareNewsweek is a Trust Project memberFor most of us, the idea that a loved one might communicate with us using AI is unsettling. After all, authenticity is a touchstone of human interactions. In the research that I conduct on community engagement, I frequently discuss authenticity and emotional intelligence—one’s capacity to read and navigate a person’s emotions. Both traits are indispensable in forging durable, trusting connections and collaborations with people, and they’re two of the few remaining things separating human cognition from that of AI.
There’s an actual name for the “ick” we feel when we suspect someone’s been using AI with us in personal discussions: Capgras syndrome. Named after French psychiatrist Joseph Capgras, the rare condition, first described in the 1920s, speaks to the deeply internalized fear we have that someone (or something) familiar is an “imposter”: in this case, your resident AI assistant.
...In recent years, AI barons have been dumping loads of R&D into addressing AI’s authenticity gap. Their hope is to eliminate AI’s creepy uncanny valley, an effect that has long made the technology feel more like sci-fi novelty than an instrument that could serve a practical function in people’s everyday lives. To this end, Big AI has been successful in, at a minimum, shaking up the notion of human authenticity in education, art and customer service. We now need to contend with the idea that AI is starting to make our relationships and personal communication to one another less authentic—and potentially less meaningful.
Our concerns on this broader matter are already high. In a March YouGov poll, 80 percent of Americans indicated being either very or somewhat concerned about AI leading to the manipulation of human behavior. That comes amid a generational erosion in trust we have in each other—only one-third of Americans in 2024 reported feeling that most people can be trusted, a decline from roughly half in 1984 when the poll’s numbers peaked. (As a snapshot, the dip in trust is about as Orwellian as could be.)
Interpersonal trust is only poised to decline as AI rapidly becomes a tool we lean more heavily on as a surrogate communicator—not just in our emails and professional reports, but also in our online dating profiles, apology letters to loved ones and birthday wishes. Increasingly sophisticated AI-generated photos and videos also mean we may be suspicious of the personal media friends and family share with us. Imagine, for instance, believing a cousin’s Instagram video of their vacation to Paris was faked via a seamless AI video generator like Sora.
While people might love using AI as a writing tool, they aren’t so keen on being on the receiving end. In studies conducted in the U.S. and U.K., researchers found that most people consider the undisclosed use of AI in interpersonal messages to be inappropriate, and when people disclose that they used AI, they tend to be trusted less.
Our best approach to dealing with AI’s infiltration into our interpersonal spaces isn’t to erect barriers to stop it. It’s to be more cognizant of when it’s used—and why. Beyond knowing the telltale signs of AI use, we need authentication tools that tell people what is and is not AI-generated. This can range from watermarking AI content, to more complex tools like cryptographic provenance tracking, which involves creating metadata, or “data trails,” that let end-users know how, when and where content was produced. But these approaches are far from infallible, with efforts to remove AI watermarks and destroy AI metadata arguably rivaling the very efforts to create them.
AI ethicists like me have long championed authentication, but lawmakers in the U.S. have yet to meaningfully support it. AI leaders like Sam Altman, CEO of OpenAI (maker of ChatGPT), have also been reticent advocates, and that’s mostly because they recognize that a key appeal of AI for many is that it feels “real.” Authentication quashes that illusion. In addition, authentication, in curtailing how we present content, raises some admittedly thorny questions about free speech. However, society’s need for transparency should outweigh this. And ultimately, deeper political commitment to authentication could help reverse the public’s broad distrust of AI companies and facilitate more comfort around AI’s integration into our lives.
In 2024, the Biden administration began requiring federal agencies to publicly disclose their AI utilization. Currently, President Donald Trump’s AI action plan, in keeping with his administration’s general antagonism toward transparency initiatives, doesn’t embrace this approach. Congress should that require AI platforms—such as those offered by OpenAI, Google, Microsoft and Apple—incorporate provenance tracking and provide users with a seamless way to determine whether a particular piece of content, such a video clip, song, or even text message, was produced by AI. Guardrails like these are imperfect, but they may make someone who’s intent on using AI in a deceitful or subversive way think twice. And these guardrails may yet help us keep at least a sliver of lives completely AI-free—assuming that’s what we actually want.
Jerel Ezell is a sociologist and assistant professor at the University of Chicago Medical Center. He studies the cultural aspects of health and technology.
The views expressed in this article are the writer's own.
Request Reprint & LicensingSubmit CorrectionView Editorial & AI Guidelines
Add Newsweek as a preferred source on Google to see more of our trusted coverage when you search.Recommended For You
OpinionConventional Wisdom: The Thanksgiving Tragedies Edition3 min readOpinionAI Is Coming for Our Most Intimate Communications. Congress Must Act | Opinion5 min read
OpinionConventional Wisdom: The Anti-Thanksgiving Edition3 min read
OpinionHow to Remain Grounded and Thankful Amid the Chaos5 min read
OpinionThis Thanksgiving, Affordability Isn’t Just About the Turkey. It’s Everyday, Every Bill, Every Choice | Opinion4 min read
OpinionINTERPOL Must Not Elect One of Its Chief Abusers As Its Next President | Opinion4 min readRelated Podcasts
Top Stories
NewsNational Guard Member Sarah Beckstrom Dead After DC Shooting3 min read
NewsUncommon Knowledge: Trump, Biden and the Real Record of Afghan Vetting5 min read
NewsZelensky’s Chief of Staff Raided As Ukraine Corruption Scandal Escalates4 min read
NewsHow a Win in Venezuela Could Be Great News for Trump5 min read
NewsMap Shows Where Significant Snow Will Disrupt Post-Thanksgiving Travel3 min read
NewsCalifornia Uber Driver Shot In Head By Passenger On Thanksgiving1 min readTrending
New York8 Million Inflation Refund Checks Sent Out Ahead of Thanksgiving4 min read
WeatherWinter Storm Map Shows Where Up to 12 Inches of Snow Could Strike3 min read
WalmartWalmart Recall Update: Customers Warned ‘Immediately Stop’ Using Products4 min read
DogsAfter 2 Days at Daycare, German Shepherd’s Return Goes Viral: ‘Never Again’3 min read
CatholicPope Leo Updates Marriage Rules for 1.4 Billion Catholics3 min readOpinion
OpinionAI Is Coming for Our Most Intimate Communications. Congress Must Act | Opinion5 min readOpinionConventional Wisdom: The Thanksgiving Tragedies Edition3 min readOpinionConventional Wisdom: The Anti-Thanksgiving Edition3 min readOpinionHow to Remain Grounded and Thankful Amid the Chaos5 min readOpinionThis Thanksgiving, Affordability Isn’t Just About the Turkey. It’s Everyday, Every Bill, Every Choice | Opinion4 min read