Detecting a fake voice is not the same as stopping a fake call
Most detection tools answer one question: is this clip synthetic? Attackers are not sending clips. They are joining calls, sitting in interviews, and asking for a wire. This is where we publish what we find: how the attacks actually run, which signals catch them, and where the tools fall short.
Monthly analysis of AI social engineering, voice fraud and deepfake attacks on enterprises.
One email a month. No spam, and we never share your address.
Latest
The newest research and analysis, updated as we publish.
Vendor Fraud: Meaning, Examples & Prevention
Vendor fraud diverts company funds through fake invoices, impersonation, and altered bank details. The main types, two real cases, and the controls…
How to Evaluate Deepfake Detection Tools: A Buyer’s Checklist
Vendor accuracy scores tell you little about production performance. Eleven criteria for evaluating deepfake detection tools, and what to test for each…
How to Prevent BEC Attacks
BEC attacks exploit trust rather than systems. Eight practical strategies to verify requests, harden email authentication, and catch AI-driven voice and video…
Emerging threats
Where the attack surface is moving next. New attacker capabilities that change what can be automated, documented before they show up in incident data.
Agentic AI Fraud: The New Attack Surface Nobody’s Talking About Yet
AI agents plan and act across systems with little human oversight. Anthropic documented a campaign in which AI executed 80 to 90% of the attack, with humans stepping in at four to six decision points. Here is how agentic fraud works, and what to verify before an agent acts.
Attack analysis
Each of these reconstructs a real attack in sequence: what the attacker had, what they said, and the exact point at which a detection layer would have changed the outcome.
Deepfake Job Interviews: Inside the Rise of AI-Generated Fake Candidates
17% of hiring managers have already interviewed a candidate using deepfake technology. Gartner forecasts one in four candidate profiles could be fake by 2028. Here is how the fraud works, and how to verify who you are hiring.
Famous Deepfake Scams: The 2024 Arup Cyber Attack
The $25M video-call fraud, reconstructed step by step.
Caller ID Spoofing: How Fraudsters Impersonate Banks
Why the number on the screen proves nothing about who is calling.
Real Estate Wire Fraud: How Closing Scams Work
The highest-value social engineering target in the economy.
No articles in this topic yet.
Tool evaluations
Vendor-neutral comparisons, including tools that compete with ours. A benchmark score is not a defence, so each entry states what the tool detects, what it does not, and the conditions under which it degrades.
Best AI Deepfake Audio Detection Tools of 2026
Ten tools sorted by the job they are built for, and why a lab-benchmark score collapses on a real phone call.
10 Best AI Deepfake Detection Tools in 2026
Voice, video and image detection compared across delivery model and coverage.
8 Best AI Social Engineering Defense Tools in 2026
What covers the full attack sequence versus a single channel.
No articles in this topic yet.
Detection playbooks
Implementation detail for teams already past the buying stage: which signals to read, which layers to stack, and what each layer still misses.
No articles in this topic yet.
Start here
Four terms get used interchangeably and mean different things. These are the plain definitions the rest of this research builds on.
No articles in this topic yet.
Browse by topic
See what your voice channel is missing
Diopter scores live voice and video for synthetic media, and for the conversation patterns that push a person toward an irreversible action.