AI, TMI? Users Tell Chatbots What They Wouldn't Want Their Closest Friends to Know

New Aura research reveals how personal disclosures can compound as AI tools remember, connect and act, shifting online risk beyond any single account or device

BOSTON, Oct. 1, 2026 /PRNewswire/ -- Aura (ASX: AXQ), a leading AI-powered online safety platform for individuals and families, today released its AI TMI Study, revealing a striking gap between awareness and behavior: Despite concerns about their identity, finances and privacy, AI users continue sharing sensitive personal information with chatbots and agents.

At the start of Cybersecurity Awareness Month, Aura is sounding the alarm on "AI TMI"—the compounding risk created when information shared with AI combines with the accounts, data and permissions it can access, adding up to "too much information" in the wrong hands. Together, these pieces of information can form a detailed, actionable profile—and a new attack surface that antiquated cybersecurity practices were not designed to protect.

"Most people think the risk is that one embarrassing prompt might leak. The bigger risk is that AI can piece together who you trust, where you bank, what frightens you and how you make decisions," said Hari Ravichandran, Founder and CEO of Aura. "In the wrong hands, that context can create a scam designed not for someone like you, but specifically for you. In a protection system, that same context can help establish what's normal and recognize when something changes. As AI gains access to accounts and the ability to act, security will need to spot when something is wrong before you do."

That is how Aura is putting AI to work: using memory and context to anticipate risk and protect families before harm occurs. Complete findings from the Aura AI TMI Study—based on an online survey of 2,000 U.S. adults ages 18–64 who use AI agents or chatbots—show why that approach is becoming increasingly urgent.

AI Has Become a Confidant

AI chatbots' conversational, one-on-one nature invites personal disclosures with the assumption of discretion.

• 73% of respondents have shared something with AI they wouldn't post publicly.
• 71% have told AI something they'd be embarrassed or uncomfortable for their closest friend or partner to know.
• 65% have entered private information about another person without first asking that person's permission.
• 72% have followed advice from an AI while making an important personal decision, including 42% who stated it significantly influenced what they did.
• 53% assume their AI conversations are completely or mostly private.

Risk Awareness Does Not Necessarily Change Behavior

Even users concerned about exposure continue sharing sensitive information.

• 86% have considered the possibility that information from their AI conversations could be exposed through an account compromise, data breach or other security incident.
• 66% are very or somewhat concerned that information shared with AI could compromise their identity, finances or privacy.
• Among those concerned respondents, 81% have continued sharing sensitive personal information with AI.

Digital Fluency Does Not Guarantee Safer AI Use 

A generational divide reveals a paradox: Across several measures, Gen Z reports riskier behaviors than Gen X, suggesting that greater comfort with these tools may also create greater exposure.

• 83% vs. 61% have shared something with AI they wouldn't post publicly.
• 79% vs. 50% have shared another person's information without permission.
• 74% vs. 51% have granted AI permissions without fully understanding them.
• 69% vs. 45% have initially believed an AI-assisted scam was legitimate.

Generational subgroup bases: Gen Z (born 1997–2008), n=350; Gen X (born 1965–1980), n=650.

The Bigger Exposure May Be What People Connect—Not What They Type

64% have given an AI-powered service access to an account or personal information without fully understanding the permissions, including:

• Email: 29%
• Personal photos or files: 24%
• Social media accounts: 23%
• Browser activity: 22%
• Text messages or messaging apps: 21%

That access can expose years of personal activity. Additionally, if users don't know what AI can reach, they can't know what could be compromised.

AI Doesn't Forget, and That Changes the Risk

When AI remembers, its purpose matters. Everyday conversations build a detailed personal profile, and the findings trace a progression from accumulated information to more persuasive scams and, ultimately, real-world harm:

• 74% have been surprised by how much an AI chatbot seemed to understand about them based on information they previously shared.
• 61% would share less personal information if an AI chatbot remembered everything they had ever told it.
• 51% say a scam using information from their AI conversations would be harder to recognize.
• 58% have initially believed an AI-generated or enhanced scam was legitimate. Among those fooled, 18% shared sensitive information and 17% lost money.
• 72% say it will be extremely or very important for security tools to detect and stop suspicious or unauthorized actions taken through AI assistants.

AI is collapsing the distance between disclosure and damage. The same context that can make an AI-powered threat more persuasive can help a protection system recognize when something is wrong. The difference is how—and why—that context is used. Security must see the risk taking shape, anticipate what comes next and intervene before exposure becomes an attack.

For more tips on how to protect your personal information, visit Aura's guide.

Talker Methodology

Talker Research surveyed 2,000 American adults between 18–64, who use AI agents or chatbots. The survey was commissioned by Aura and administered and conducted online by Talker Research between September 3 to September 10, 2026. To view the complete methodology as part of AAPOR's Transparency Initiative, please visit the Talker Research Process and Methodology page.

About Aura

Aura (ASX: AXQ) is a global leader in online safety and wellbeing. Built on the belief that people deserve a trusted, always-on layer of protection, Aura's AI-powered platform delivers protection for individuals, families, and enterprises—from proactive protection against identity theft, financial fraud and online threats to tools that help schools and parents protect children from cyberbullying, harmful content, and threats to their wellbeing.

Aura's platform spans the environments that matter most—home, school, and work—empowering people of all ages with end-to-end protection for every aspect of online life. Learn more at aura.com

SOURCE Aura