
Social engineering is the art of manipulating a person into giving up information, access, or an action they would not grant if they thought it through. The attacker does not break a system: they convince someone to open the door. It can happen by email, text, phone call, or in person.
This article covers the definition, the main techniques, example scenarios, the psychological levers behind them, the signs that should make you stop, and how to train a team. It extends our guide to security awareness training and vishing practice.
NIST defines it as the act of deceiving an individual into revealing sensitive information, obtaining unauthorized access, or committing fraud, by gaining the person's confidence and trust. CISA, the US cybersecurity agency, describes an attack in which the attacker uses human interaction to obtain or compromise information about an organization.
The word “engineering” can mislead: nothing about the target is technical. What gets attacked is a human decision, made fast, under pressure, with good intentions.
That is why these attacks work. According to the 2025 activity report of Cybermalveillance.gouv.fr, the French national cyber victim assistance service, more than 500,000 requests for assistance were recorded, up 20% from 2024. A firewall does not protect against an employee who reads out a code in good faith.
Techniques differ by channel and by pretext. The principle stays the same: build a situation in which the person complies.
The three situations below are illustrative scenarios built for this article. They do not describe real cases.
Imagine a call to an executive assistant. A man introduces himself as IT support and reports an alert on her computer. To “fix” it, he asks for her password, then for the code she just received by text. The pretext is plausible, the department really exists, and the help was offered without her asking.
A treasury manager gets a message, then a call, from the “CFO.” A confidential acquisition requires a wire transfer within the hour, and he must not tell anyone. Every element discourages checking: authority, urgency, secrecy.
A small business owner is called by a supposed advisor from her bank. A suspicious transaction is in progress, it must be blocked now, and to do that she has to confirm a code received by text. The code blocks nothing: it approves the attacker's transfer. Our article on the fake bank advisor scam method walks through this sequence step by step.
Psychologist Robert Cialdini describes in Influence six principles that steer our decisions: reciprocity, commitment and consistency, social proof, authority, liking, and scarcity. Fraudsters apply them to a short conversation, with a script written in advance. On the phone, six levers come up almost every time.
A fraudulent call often combines three or four of them in under two minutes. Naming the levers in your head, one by one, breaks the pressure.
Whatever the channel, a few signs come back. None is enough alone, but two or three together justify stopping.
When in doubt, CISA recommends contacting the organization directly using contact details from a previous statement, never those in the message. That is the idea behind the callback through an independent channel. For the questions to ask before calling back, see how to verify a caller's identity.
Knowing the list of techniques helps. It does not guarantee you will spot them live: on the phone, the brain is busy with the conversation, the wish to be polite, and the fear of getting it wrong. The easiest behavior, which is to comply, takes over.
A reflex is built like any skill: by repeating it in conditions close to the real thing. That is the principle of role-play. The learner speaks out loud, feels the pressure, tries a phrasing, fails, and starts again. Our article on how to say no on the phone under pressure covers the phrasings that hold.
With Face Up, a trainer writes a scenario from a real call experienced in their organization. A persona played by an AI takes the caller's role, chains the levers, and reacts to the learner's answers, by voice or video. After the session, a debrief shows what held, what gave way, and quotes the phrases spoken. The learner can replay. To understand the principle, see our explanation of AI role-play.
It is the set of manipulation techniques that push a person to disclose information, grant access, or take an action, by exploiting trust and pressure rather than a technical flaw.
No. It happens by email, text, messaging apps, and in person (tailgating, fake contractors). The phone gives it an edge, though: a voice inspires trust, and the pace of the exchange prevents checking.
Social engineering is the family, and phishing is one of its techniques. Phishing uses a written message or a fake site, vishing uses voice, and smishing uses text messages.
Act right away: alert the relevant team, block whatever can be blocked, and note the details of the exchange. Feeling guilty must not delay the report. Write down the time, the number shown, and the phrases you heard.
To train your teams to spot these techniques out loud, explore Face Up role-plays or book a demo with our team.
From reading to practice.
Pick a real-life situation and talk to an AI persona, by video with an avatar or by audio call. 5 to 10 minutes, no account, instant debrief.
Try a scenario →