Most People Lie Less to AI Than to Their Friends

31 May 2025

Our interactions are increasingly mediated through non-human agents, chatbots, virtual assistants, AI therapists, and customer service bots. Surprisingly, research and anecdotal evidence suggest a counterintuitive trend: people lie less to AI than they do to their friends.

This statement opens up a fascinating window into human behavior, trust, social dynamics, and the evolving role of technology in our lives. Why are we more honest with machines than with people who know and love us? What does this say about our psychology, and what are the implications for AI developers, society, and interpersonal communication?

This article delves deeply into the reasons behind this phenomenon, supported by research, psychological theories, and technological insights.

The Human Tendency to Lie

Before exploring why people may lie less to AI, it's crucial to understand why humans lie in the first place. Lying is a common social behavior, often motivated by:

Protecting oneself from consequences
Avoiding hurting others' feelings
Preserving self-image
Manipulating situations to one's advantage
Conforming to social expectations

According to various studies, the average person tells one to two lies per day, often without malicious intent. These lies range from harmless “white lies” (“You look great in that outfit”) to more serious deceptions.

When interacting with friends or loved ones, people often lie to preserve relationships or avoid conflict. These lies are socially strategic.

Why People Are More Honest with AI

1. Lack of Judgment

AI doesn’t judge. Unlike friends, AI doesn’t react emotionally, hold grudges, or change its opinion of you. This creates a non-threatening environment where people feel safer revealing the truth.

A 2021 study from the University of Oxford showed that participants were more likely to disclose sensitive personal information to a chatbot than to a human interviewer.

2. Anonymity and Distance

Many AI systems operate in asynchronous or text-based formats that provide psychological distance. People are more candid when they feel anonymous, as there's no immediate social repercussion.

In therapy or healthcare settings, AI chatbots like Woebot or Wysa have been found to elicit greater emotional openness than human therapists in early sessions.

3. Perceived Objectivity of AI

AI is seen as neutral, especially in data-driven tasks. People may believe that there is no point in lying to a system that might already “know the truth” or can verify facts easily.

For example, users filling out forms with AI-based health assessments are more likely to report accurate symptoms than when speaking to a doctor they fear might judge them.

4. No Need for Social Performance

In social settings, humans are often “performing”, managing impressions, using tone and gestures, and monitoring reactions. AI removes this performative element, enabling more direct and unfiltered responses.

This is particularly true in AI-mediated interviews, where job applicants feel less pressure to conform to expected norms.

5. Built-in Confidentiality

AI doesn’t gossip. When people talk to friends, there's always a fear, rational or not, that secrets might be shared. With AI (assuming privacy policies are followed), the perception is that the data stays within a closed loop.

Situational Honesty: Context Matters

It’s worth noting that context heavily influences honesty levels. While people may be more truthful with AI in sensitive domains (mental health, sexual health, financial concerns), they may still deceive AI in transactional or adversarial settings, like:

Gaming AI for financial benefit (e.g., lying on loan applications)
Exaggerating qualifications in AI-driven job applications
Using deceptive tactics in AI surveillance systems

In other words, people are honest with AI when it serves their interest to be honest, just as with humans. The difference lies in perceived consequences and power dynamics.

Cultural and Demographic Differences

Honesty toward AI can vary across cultures and demographics:

Younger generations, who grew up interacting with technology, are more comfortable sharing personal data with AI.
Collectivist cultures may be more reserved in AI disclosures compared to individualistic cultures that value openness.
Education level and tech literacy also play roles; people who understand AI systems may be more strategic in their disclosures.

Ethical and Design Implications for AI Developers

This trend toward honesty with AI poses both opportunities and ethical challenges:

Opportunities

Improved accuracy in health, therapy, and legal tech applications.

Better data for personalization and machine learning.

Enhanced user experience and trust.

Challenges

Risk of over-disclosure or misplaced trust in systems with unclear data policies.

Need for transparent AI design to ensure people understand how their data is used.

Ethical responsibility to protect vulnerable users (e.g., children, mentally ill individuals) from over-reliance on AI.

Developers must ensure their systems are ethically sound, data-protective, and transparent in functionality. AI should never exploit human honesty.

What This Reveals About Human Nature

The fact that people lie less to AI suggests several truths about us:

We crave nonjudgmental spaces to be honest.

Social relationships come with complex emotional negotiations.

We’re pragmatic in how we distribute truth.

It also reveals a potential loneliness or mistrust in human connections, where we prefer digital listeners over flawed human ones.

The Irony of Human-AI Honesty

It’s a fascinating irony: as artificial intelligence becomes more human-like, people become more human in return, not by hiding, but by opening up. When faced with an entity that doesn’t judge, doesn’t remember mistakes emotionally, and doesn’t have social expectations, many of us let down our guard and speak more freely.

Yet this honesty shouldn’t lead us to abandon human relationships. Rather, it should inspire us to create more judgment-free spaces in our real-world interactions.

AI, in this case, holds up a mirror to our social selves, and what we see is a desire for honesty, safety, and understanding.

References and Further Reading

Hancock, J. T., & Toma, C. L. (2009). “Lying for Love: The Ethics and Psychology of Online Dating Deception.”
Lucas, G. M., et al. (2014). “It’s only a computer: Virtual humans increase willingness to disclose.” Computers in Human Behavior.
Oxford Internet Institute. (2021). "Digital Honesty: Patterns of Truth-Telling in AI vs. Human Interaction."
Fitzpatrick, K. K., Darcy, A., & Vierhile, M. (2017). "Delivering Cognitive Behavioral Therapy to Young Adults With Symptoms of Depression and Anxiety Using a Fully Automated Conversational Agent (Woebot)."

Resources To Read