Can AI Catch a Liar on WhatsApp?
Imagine receiving a WhatsApp message from someone who says, “Baby, I swear I was sleeping all night.” But something about the story doesn't make sense.
You copy the conversation into an AI chatbot and ask one question: “Is this person lying to me?”
Within seconds, the AI may analyse the messages, identify contradictions and explain why the story seems suspicious. But does that mean it has caught the person lying?
Not necessarily.
AI tools such as ChatGPT and Gemini can analyse written conversations, compare statements and highlight inconsistencies. However, they cannot reliably determine whether someone is lying just by reading their WhatsApp messages.
How Can AI Analyse WhatsApp Messages?
AI chatbots can examine a conversation and help you notice details you might have missed. Here are three ways they can help.
1. Finding contradictions
Suppose someone tells you they were at home on Saturday but later says they spent the entire evening at a friend's house.
AI can compare the two statements and point out the inconsistency. You can then ask the person for clarification.
However, a contradiction doesn't automatically mean someone is lying. They might have made a mistake, forgotten something or explained themselves poorly.
2. Identifying suspicious requests
AI can help you examine messages that contain pressure, threats, emotional manipulation or requests for money.
For example, someone you met online might claim to be stranded and demand urgent financial assistance. AI can help you identify common scam warning signs and suggest questions to ask.
This doesn't prove that the person is a scammer, but it can help you assess the situation more carefully.
3. Comparing different versions of a story
If someone gives you two different explanations about the same event, AI can compare both versions and show you what changed.
This can be useful when checking information, reviewing business conversations or trying to understand a confusing discussion.
Still, AI cannot know what actually happened unless there is reliable evidence to support the conclusion.
I Tried These AI Prompts for WhatsApp Conversations
You can try the following prompts with an AI chatbot. Remove names, phone numbers, addresses and other private information before sharing any messages.
Prompt 1: Find contradictions
“Analyse this conversation and identify any statements that appear to contradict each other. Explain each inconsistency and suggest reasonable alternative explanations. Do not automatically assume the person is lying.”
Prompt 2: Check for possible scams
“Review these messages for common scam warning signs, including urgent requests for money, suspicious links, impersonation and emotional pressure. Explain which details are concerning and what evidence would help verify the situation.”
Prompt 3: Compare two stories
“Compare these two statements about the same event. Identify the differences, explain which details need clarification and suggest neutral questions I can ask. Do not claim that either person is lying without evidence.”
These prompts can help you understand a conversation, but their answers should be treated as suggestions rather than proof.
Can ChatGPT or Gemini Tell If Your Partner Is Cheating?
This is where things become complicated.
You could give an AI chatbot a conversation with your partner and ask whether the messages suggest cheating. The AI might highlight unusual explanations, changes in a story or statements that appear inconsistent.
But it cannot reliably determine whether your partner is cheating from text messages alone.
A short reply, a change in texting habits or an unexplained delay could have many reasons. Treating these details as proof could create unnecessary suspicion and damage a relationship.
AI can help you organise your thoughts and prepare questions for an honest conversation. It should not make the final decision about someone's honesty.
What Does Research Say About AI Lie Detection?
Research suggests that AI is not yet a dependable lie detector.
A study published in the Journal of Communication in 2025 examined AI judgements of human truthfulness across 12 experiments. The researchers found that AI's performance varied by context and that some systems showed a strong tendency to label people as liars. The findings raised concerns about relying on AI to detect deception.
The research did not establish that AI can reliably identify lies in ordinary WhatsApp conversations. Instead, it highlights why AI-generated judgements should be treated cautiously.
Important Warning Before You Upload WhatsApp Chats
Your conversations may contain private information about you and other people. Before sharing messages with an AI tool:
- Remove names, phone numbers, addresses and financial details.
- Avoid uploading intimate photos, private documents or sensitive conversations.
- Never use an AI-generated accusation as your only evidence.
- Verify suspicious links, claims and payment requests independently.
- Remember that a confident AI answer can still be wrong.
So, Can AI Really Expose Someone Lying on WhatsApp?
AI can help you identify contradictions, examine suspicious requests and compare different versions of a story. What it cannot reliably do is look at a message and know whether the sender is telling the truth.
The smartest approach is to use AI as an assistant for analysing information, not as a machine that decides who is guilty.
Next time a WhatsApp conversation feels suspicious, you can ask AI to help you examine the evidence. Just remember: spotting an inconsistency is not the same as catching a liar.