GenAI is accelerating email social engineering, not replacing it

At a glance

What this is: NHI Mgmt Group's analysis, drawing on Barracuda's report: Email-based social engineering is intensifying as conversation hijacking, QR attacks and BEC remain persistent and measurable attack patterns.

Why it matters: Email remains a primary identity attack surface, so practitioners need controls that address thread compromise, link obfuscation and mailbox-targeted fraud as one governance problem.

By the numbers:

• Barracuda reports a 70% increase in conversation hijacking since 2022.
• Barracuda found that 1 in 20 mailboxes were targeted with QR code attacks in the last quarter of 2023.
• Barracuda says 1 in every 10 attacks were business email compromise.
• Barracuda found that 40% of social engineering attacks including a shortened URL used bit.ly.

Context

Email attacks keep evolving because the control problem is no longer only malware delivery. Social engineering now blends identity impersonation, thread hijacking, shortened links and QR codes, which means defenders must treat the mailbox as a trust boundary rather than a simple message channel.

Generative AI lowers the effort needed to personalise lures and sustain believable conversation flows, so the risk shifts from obvious phishing to harder-to-see manipulation inside existing business relationships. That makes the issue relevant to both human identity governance and the protection of accounts that carry authority in email workflows.

Key questions

Q: How do I detect conversation hijacking in email threads?

A: Detect conversation hijacking by looking for mailbox access patterns that do not match the usual sender, device, location, or forwarding behaviour, then verify whether a reply came from a compromised account or a lookalike address. Mail trace logs, unusual OAuth consent grants, new inbox rules, and sudden changes in reply timing are common signals. Conditional access and mailbox auditing help confirm the abuse path.

Q: Why does BEC keep working against organisations?

A: BEC keeps working because attackers exploit process trust, not malware. They imitate executives, suppliers, or finance staff, then use stolen email accounts, lookalike domains, or reply-chain manipulation to request payments, change bank details, or divert invoices. Out-of-band verification, payment call-back controls, and enforced approval steps reduce the chance that a single convincing message is enough.

Q: How can QR codes be abused in phishing campaigns?

A: QR codes in phishing campaigns can hide the destination and bypass simple URL filters, because the user scans a code that resolves to a malicious site or login page. Destination inspection checks the encoded URL, redirects, and final landing page before access is granted. Attackers also use QR codes in attachments, posters, or emails to move the victim off the protected message path.

Q: How should organisations verify high-risk requests that arrive by email?

A: Use an independent confirmation step outside the email thread for payments, vendor changes, payroll updates, and other high-impact actions. Verification should check the requester through a separate trusted channel and confirm the business event before any action is taken.

Technical breakdown

Conversation hijacking as a thread-integrity failure

Conversation hijacking works when an attacker inserts themselves into an existing email thread or replies from a compromised account so the message inherits trust from prior correspondence. The technique is effective because recipients tend to validate context, not origin, once a thread already looks familiar. Generative AI can make the wording more natural, but the core problem is still trust transfer inside a live conversation. Detection therefore depends on behavioural anomalies, reply-path changes and sender identity inconsistency rather than content scanning alone. Read Top 10 Email Security Threats.

Practical implication: Monitor thread continuity, sender drift and anomalous reply behaviour to catch business email compromise before payment or data disclosure occurs.

QR code attacks and link obfuscation in email

QR attacks shift the malicious payload out of the visible email body and into a machine-readable image, which can bypass some text-based filtering and user scrutiny. Once scanned, the QR code can route the user to credential harvesters, consent scams or other socially engineered destinations. Shortened URLs play a similar role by hiding the final destination and making reputation checks harder. The defensive gap is not only scanning, but also validating the destination and context before the user leaves the mail client. See the QR Code Phishing (Quishing) Guide.

Practical implication: Inspect QR payloads and shortened links with the same policy rigor as ordinary URLs, because obfuscation is part of the attack path.

Why BEC remains a high-value email fraud pattern

Business email compromise remains attractive because it exploits authority, urgency and routine financial processes rather than technical vulnerabilities. Attackers need only enough access to impersonate an executive, supplier or internal approver at the right moment. The data in this report shows that BEC still occupies a large share of social engineering volume, which suggests organisations are not only fighting phishing, but also weak verification around payment and request changes. The real exposure is process trust, not just inbox hygiene.

Practical implication: Tie payment, banking-detail changes and approval requests to out-of-band verification so email alone cannot authorise money movement.

Threat narrative

Attacker objective: The attacker wants to turn ordinary business email trust into fraudulent access, payment diversion or other monetisable account abuse.

1. Entry begins with a lure delivered through email, often using a shortened URL, QR code or trusted-looking conversation context to pull the victim into the interaction.
2. Credential_harvested or trust-abused when the target follows the link or replies within the thread, giving the attacker either account access or believable access to an existing business process.
3. Escalation occurs as the attacker uses the trusted mailbox or impersonated identity to extend the fraud, hijack the conversation or steer the recipient toward payment or data disclosure.
4. Impact is financial theft, business process manipulation or broader compromise of the mailbox trust relationship that supports future fraud.

NHI Mgmt Group analysis

Email fraud is moving from message delivery to relationship abuse. The report's most important signal is not a single lure type, but the way attackers now exploit established communication context to get past user judgment. That changes email security from a filter problem into a trust-governance problem, especially where mailbox access intersects with finance, procurement and executive authority. Practitioners should treat thread integrity as a control objective, not an email convenience.

Conversation hijacking creates a named control gap: thread trust without identity revalidation. Once a sender appears inside an existing exchange, the recipient often stops checking for origin and starts checking for relevance. That assumption fails when attackers can reuse context, copy tone and time messages to business routines. The useful response is to make high-risk requests re-verified at the point of action, not merely at inbox entry.

QR codes and shortened links show how social engineering now combines human error with machine-mediated routing. A QR code moves the risky hop away from the visible message, while a shortened URL hides destination and intent until after a click. That is why email defence has to extend into destination inspection, policy enforcement and user verification before the browser opens. Security teams should measure whether their controls still work once the payload leaves plain text.

Gmail-based BEC illustrates that sender reputation alone no longer defines trust. When a familiar platform becomes part of the attack path, allowlisting and brand familiarity can mislead reviewers more than help them. The field needs a stronger model of authenticated intent, where payment changes and account updates require independent confirmation. Email governance should assume that recognizable infrastructure can still carry fraudulent requests.

Conversation integrity debt: organisations that rely on historic thread context, sender familiarity and link reputation are accumulating a hidden exposure that attackers can amortise across multiple messages. The report shows that the threat is not a one-off phishing spike, but a durable business process weakness. Practitioners should move from inbox-centric controls to workflow-aware verification.

From our research library:

• Generative AI use specifically increased from 33% in 2023 to 79% in 2025, according to McKinsey’s Global Surveys on the State of AI.

What this signals

Mailbox trust has become an operational risk signal, not just a security alert. Email programmes now need to watch for where a request originated, how a thread evolved and whether a message is trying to inherit credibility from prior correspondence. Controls that only score the sender or URL will miss the abuse of business context that makes modern social engineering effective.

Thread-aware verification will matter more as AI-assisted lures improve. The challenge is not to recognise every synthetic phrase, but to ensure that high-risk actions cannot be completed through a single compromised exchange. Practitioners should design approval paths that survive realistic impersonation and conversation reuse.

For practitioners

• Harden conversation-based approvals Require out-of-band verification for payment, banking detail and beneficiary changes even when the request arrives in an existing thread.
• Treat QR codes as hostile payloads Route QR scans through secure inspection and user warning layers so the destination is checked before the browser opens.
• Correlate thread anomalies with identity signals Flag sudden sender changes, unusual reply timing and mailbox access from unfamiliar locations as indicators of conversation hijacking.
• Review shortened-link policy in email flow Block or detonate shortened URLs in business email and require destination expansion before users can reach external content.

Key takeaways

• Email social engineering is increasingly built around trust reuse inside live conversations, not only around malicious attachments or obvious phishing pages.
• Barracuda's data shows measurable pressure points in conversation hijacking, QR-based lures and BEC, which means these attack styles are still operationally effective.
• The control answer is to move verification closer to the transaction, with thread-aware detection, destination inspection and out-of-band approval for risky requests.

Related NHI Mgmt Group resources: Top 10 Email Security Threats · QR Code Phishing (Quishing) Guide · Breach and Attack Simulation (BAS) Guide

Key terms

• Social Engineering: Social engineering is the use of deception, urgency, and authority to persuade a person to reveal information or take a risky action. It targets human decision-making rather than software defects, and often turns legitimate identity workflows into the attack path.
• Conversation Hijacking: Conversation hijacking is the insertion of a fraudulent actor into an existing email thread so the request appears to continue a legitimate discussion. The attacker relies on prior message history, familiar tone, and trusted recipients to bypass human suspicion and complete the fraud path.
• Business email compromise: A form of social engineering where an attacker impersonates a trusted person or domain to manipulate payment, change banking details, or extract sensitive information. It often succeeds without malware because the attacker targets process trust and human judgement instead of technical controls.
• QR Code Phishing: A phishing method that hides a malicious destination inside a QR image instead of a visible link. The code is scanned by a user device, which can move the victim into a login page, credential prompt, or session capture flow that bypasses simpler link-based inspection.

Deepen your knowledge

NHI Mgmt Group covers identity security, NHI governance, and agentic AI through independent research, practitioner guides, and the NHI Foundation Level course. Explore nhimg.org for resources that connect identity governance to the broader security disciplines your programme depends on.