Voice Fraud: Why Banks Can't Trust Callers Anymore | Anat Goldstein, Founder of FinOptima

For centuries, a voice was proof. You heard your banker, your child, your CEO, and you knew. Not anymore. A voice can now be cloned from a few seconds of public audio. The problem is not that voice disappears, it is that voice becomes easy to manufacture.

This week I speak with Anat Goldstein, Founder of FinOptima, whose whole argument is that no single signal, not a voice, not a password, not a device, can prove who is on the other end of a call anymore.

FinOptima builds explainable fraud detection for the institutions the industry skipped: community banks and credit unions, which face the same deepfake threats as the top 20 banks on a fraction of the budget and with no in-house fraud team. The numbers are not abstract. Fraud losses are projected at 40 billion dollars. In one case, criminals deepfaked senior executives on a video call and moved around 25 million, and the employee who sent it was not careless.

We get into why fraud now happens inside the live call rather than before it, what regulators mean when they demand a reason for every blocked transaction, and why the harder problem coming is not detecting fakes but proving which interactions are real.

Anat also traces the move from a 20-year banking career to founding a company, validated through customer discovery before any code, and built one hard decision at a time.

What you'll learn:

• Why a voice can no longer prove identity
• Why one signal only creates suspicion, and how voice, behavior, device and transaction history together are what create confidence
• Why fraud detection has to happen during the call, not after the money has moved
• Why community banks and credit unions carry the same threat as the largest banks with almost none of the resources
• When it makes sense to partner rather than build fraud infrastructure in-house
• What "explainable" fraud AI actually means to a regulator, and why a risk score on its own no longer holds up
• What fraud looks like once AI makes it cheap enough for anyone to run, and why trust infrastructure has to come first
• How Anat found and validated the idea through customer discovery, and what she means by resilience built one decision at a time

Timestamps

00:00 — Cold open: your voice is no longer proof

02:24 — 40 billion in losses, and a 25 million dollar deepfake call

04:13 — How do you know the caller is real?

05:21 — Why community banks and credit unions are the target

07:09 — Why today's fraud tools miss it: they work in silos

08:10 — Fraud that happens inside the live call

08:59 — What explainability means to a regulator

12:08 — One signal creates suspicion, many create confidence

17:17 — Build versus buy, and the first 90 days

20:16 — Leaving a 20-year banking career to found a company

24:37 — Resilience, built one decision at a time

27:50 — When fraud gets cheap: good AI versus bad AI

Connect with Anat:

🔗 FinOptima: https://www.linkedin.com/company/finoptima-solutions/

🔗 LinkedIn: https://www.linkedin.com/in/anat-goldstein/

Connect with Monica

🔗 LinkedIn: https://www.linkedin.com/in/monicamillares/

🔗 Purpose Driven FinTech: https://www.youtube.com/@monica_millares/videos

🔗 Purpose Driven FinTech. Spanish: https://www.youtube.com/@MonicaMillares_Español/videos

Disclaimer: This episode does not constitute professional nor financial advice and does not represent the opinion nor views of my current, past or future employers. The guest has agreed to record and release our conversation for the use of this podcast and promotion in social media.