Key facts
- Live social engineering campaigns employ deepfakes, helpdesk impersonation and fake IT support requests to trick employees and obtain unauthorized entry to corporate systems.
- A finance employee at Arup's Hong Kong office got an email in January 2024 that seemed to come from the UK-based CFO and asked for several transactions.
- Though initially doubtful, the employee accepted the request as real after seeing a video meeting that seemed to show the CFO and coworkers.
- Multiple transfers followed, and the loss totaled approximately HK$200 million (US$25.6 million).
Live social engineering campaigns employ deepfakes, helpdesk impersonation and fake IT-support requests to trick employees and obtain unauthorized entry to corporate systems, BleepingComputer reported.
In January 2024, a finance employee at Arup’s Hong Kong office received an email that appeared to come from the UK-based CFO and asked for several transactions. Though initially doubtful, the employee accepted the request after viewing a video meeting that seemed to show the CFO and coworkers. Hong Kong police said the attackers built the convincing pre-recorded conference from publicly available video and audio. Multiple transfers followed, and the loss totaled approximately HK$200 million, or US$25.6 million.
Marks & Spencer said a major cyber attack in April 2025 disrupted its operations for weeks. The attack suspended online orders, affected contactless payments and Click & Collect services, and resulted in stolen customer data. Before mitigation, the retailer estimated that the incident would lower its operating profit for 2025/26 by about ยฃ300 million. Investigators tied the breach to Scattered Spider, a cybercriminal group that uses social engineering to enter corporate networks, and Marks & Spencer later confirmed that deceiving employees at a third-party contractor, not exploiting a technical vulnerability, gave attackers entry.
Microsoft described a campaign where threat actors pretended to be IT support staff on Microsoft Teams and convinced employees to grant remote access to their devices. Using external Teams messages or calls, the attackers posed as internal IT personnel. With pretexts such as urgent technical issues, they persuaded employees to allow remote control through screen sharing or legitimate support tools including Quick Assist. After connecting, the attackers used PowerShell to install malicious software, create persistent access and collect information about the compromised network, and Microsoft additionally observed lateral movement attempts aimed at domain controllers and certificate authorities.
BleepingComputer reported that social engineering has long been a key technique and that an organization can be compromised without sophisticated malware when an attacker convinces someone they are a trusted colleague. Verizon’s Data Breach Investigation Report found that stolen credentials play a role in 44.7% of breaches.
BleepingComputer noted that deepfake technology makes impersonation especially convincing because familiar faces and voices can be recreated, and that attackers do not need AI to use believable pretexts and trusted communication channels. Before attacking, they conduct reconnaissance to collect information that makes their social engineering attempts seem like routine requests. Routine activities such as password resets and remote IT support can let attackers gain access when requests are not properly verified. Third-party providers increase the attack surface because a company can tightly manage its own systems yet lack insight into how outside support teams confirm identities and approve sensitive actions.
BleepingComputer reported that security awareness training helps employees spot suspicious requests, yet well-trained staff can still be fooled by convincing impersonation attempts. Because social engineering tactics are becoming more convincing, organizations cannot rely on employees to recognize every fraudulent request. IT and security teams must determine how to verify that a person requesting access or approving a sensitive action is legitimate.

BleepingComputer reported that reliable verification processes should support employees who must make security-sensitive decisions instead of requiring them to judge whether someone is trustworthy. Service desk teams should verify callers with trusted authentication methods instead of relying on security questions or personal details attackers may possess. Organizations can use Specops Secure Service Desk to add verification to their service desk processes.
BleepingComputer reported that agents can verify callers with more than 15 multifactor authentication factors, including choices that work without a phone. Duo and Okta are among the existing identity platforms that the tool supports. The Specops Authentication API enables secure workforce authentication and identity verification within existing applications, automated processes and business workflows.
BleepingComputer reported that traditional support channels define the verification standard that internal chatbots, self-service portals and automated IT processes must also meet. Strong identity verification should be required before processing a password reset request that comes through an internal chatbot. Any financial transaction, privileged access or other sensitive request must first receive independent confirmation through trusted channels.
BleepingComputer reported that one option is to contact the requester through previously verified details, and another is to add extra authentication to approval workflows. Additional identity verification can help ensure that authorized users make sensitive requests when support processes run through internal applications or automated workflows.
BleepingComputer reported that broader security controls such as phishing-resistant multifactor authentication and least-privilege access should support these measures, along with monitoring for suspicious account activity. These defensive steps can lower risk as social engineering tactics grow more convincing. When checks are embedded in service desks and applications as well as everyday workflows, attackers cannot easily turn a convincing impersonation into an actual compromise.

Join the discussion