40+ startups calling BS on AI fakes

Meet Ervin Hoskins. He wrote op-eds about the Iran war. A small Florida newspaper ran one.

Small problem. Ervin isn’t real. Last week, OpenAI said he was one of seven fake journalists run out of Iran, with ChatGPT polishing the copy.

AI is faking voices, faces, and job candidates, too. We’ve known this for a while, but it just keeps getting more dystopian. 

So we went digging and mapped the 42 VC-backed companies in the AI fake-catching business to see where the market stands.

This brief is adapted from our weekly newsletter. Sign up here to get it in your inbox every Thursday.

Four things stand out:

1. Fraud teams are the buyers

Companies that already verify identities and investigate claims are bringing in outside detection.

• Rival credit bureaus Equifax and Experian both license Incode, which verifies identities and detects deepfakes and synthetic identities. Incode has 75+ partners and customers in our data, and claims to serve 8 of the 10 largest US banks.
• Truepic, which verifies photos and video for insurance and lending claims, is now built into Verisk’s claims software. It also previously partnered with Dun & Bradstreet.

That’s also where the money is: since January 2025, the companies on our map have raised about $125M across 26 deals, and fraud fighters catching fake voices and identities took $74M of it. Incode raised the largest round at $37M.

Banks are also investing: Capital One backed deepfake detector GetReal and BNY backed Reality Defender in 2025.

2. New founders are chasing fraud, too

Of the 11 companies on our map founded since 2023, 7 sell mainly to banks, insurers, or teams fighting impersonation on calls and video meetings. These include:

• Neural Defend, founded in 2024, detects deepfaked video and audio in real time for financial firms.
• DeepXL, also founded in 2024, sells a deepfake detection API to financial services companies and online platforms.

Few of these founders come from fraud teams directly. For instance, IdentifAI’s founder ran Italian cybersecurity firm Yoroi for nine years, and DeepXL’s CEO was a lawyer at Nordic banks Danske Bank and Handelsbanken.

And 6 of the 7 fraud-focused startups have at least one technical founder, bringing engineering experience from fields like self-driving cars and security software.

3. Detection is turning into a feature

Software companies are building detection into products their customers already use, many through partnerships and licensing deals.

• Substack added a button that lets readers scan posts with AI-text detector Pangram.
• Zoom built voice-fraud detector Pindrop into its contact center software, and FICO added Pindrop to its marketplace in May. 
• In March, Charm Security gave its AI fraud agents the ability to call Reality Defender’s deepfake detection directly, so teams don’t have to switch tools. 

4. The next deals are lining up

Expect Equifax to be on the hunt. It already runs Incode inside Kount, the fraud platform it acquired, and on its July earnings call it named identity and fraud as one of its four acquisition lanes.

On the other side of the table, the likeliest targets are small companies built for fraud and identity teams, and several are due for fresh capital. Four on our map are in their predicted funding windows right now:

• IdentifAI: spots deepfaked images, video, and voices; partners with insurer AmTrust
• DeepTrust: catches deepfaked voices on calls and video communication
• Scam AI: detects AI fakes for financial services, contact centers, and hiring teams
• AI or Not: flags AI-generated media for fraud and identity checks

As for Ervin, nobody’s bidding on him. Fake columnists are cheap to make and hard to sell.

For information on reprint rights or other inquiries, please contact reprints@cbinsights.com.

If you aren’t already a client, sign up for a free trial to learn more about our platform.