As reported by BleepingComputer, the convergence of deepfake technology and traditional social engineering is creating a new class of identity verification challenges that most organizations are fundamentally unprepared to address. The Arup case—where a finance employee transferred $25.6 million after participating in a fabricated video conference—and the M&S breach attributed to Scattered Spider via third-party helpdesk deception, are not isolated incidents. They represent a systemic weakness in how organizations verify identity during high-stakes interactions.
The Threat Surface Has Shifted
For decades, security teams have focused on technical controls—patching vulnerabilities, hardening network perimeters, deploying EDR. But the attack surface that matters most in 2026 is human trust, and adversaries are weaponizing AI to exploit it at scale. The barrier to producing convincing deepfakes has collapsed. Open-source models can clone a voice from seconds of audio, and face-swap technology now runs on consumer hardware with near-real-time latency.
The question is no longer 'can attackers fake identity?' but 'can your verification process distinguish between real and synthetic identity under pressure?'
Three Attack Patterns Converging
Why Traditional MFA Falls Short
Multi-factor authentication, while essential, was designed to verify that a credential holder has possession of a second factor—not to verify that the person requesting an action is genuinely authorized to do so. When an attacker convinces an authenticated employee to perform an action on their behalf, MFA provides no protection. This is the verification gap: the space between authentication (who holds the credential) and authorization intent (who genuinely wants this action performed).
Who Is Most at Risk
- Finance and engineering firms with high-value transaction workflows and geographically distributed leadership
- Organizations with large third-party contractor ecosystems where helpdesk verification standards vary by vendor
- Companies using collaboration platforms (Teams, Slack, Zoom) for internal and cross-organizational communication without identity verification controls
- Mid-market enterprises lacking dedicated threat intelligence teams to track groups like Scattered Spider
Shield53 Recommendations
Immediate Actions
- Implement out-of-band verification for high-value transactions: Any financial transfer above a defined threshold must be verified through a separate, pre-established channel (e.g., a phone call to a known number, not one provided in the request).
- Deploy liveness detection for video-based authorization: If video confirmation is required for approvals, use platforms with active liveness checks rather than passive video streams that can be pre-recorded or deepfaked.
- Restrict external collaboration tool access: Configure Teams and similar platforms to restrict external federation, require admin approval for external participants, and flag messages from unverified external senders.
- Audit third-party helpdesk procedures: Require all vendors with network or system access to document and demonstrate their identity verification protocols. Make this a contractual obligation.
Strategic Initiatives
- Adopt continuous identity assurance: Move toward risk-based, contextual authentication that evaluates session behavior, device posture, and network origin—not just credential validity.
- Train for synthetic media scenarios: Update security awareness training to include deepfake recognition and the principle that video presence does not constitute identity proof.
- Establish transaction confirmation protocols: Define a tiered system where higher-risk actions require escalating verification—dual authorization, time-delayed execution, or cryptographic signing.
- Monitor for Scattered Spider TTPs: This group specifically targets identity systems and helpdesk workflows. Track their known indicators and social engineering patterns through threat intelligence feeds.
The organizations that survive this next wave of social engineering will be those that treat identity verification as a dynamic, risk-adjusted process rather than a binary checkpoint. Authentication confirms a credential; verification must confirm intent, context, and authorization—every single time.