Siri co-founder backs Helix's on-device voice ID as CAPTCHA alternative

Helix raises $6 million for voice-based verification

Helix, a startup founded in 2025, has raised $6 million to build voice-based identity verification that can replace or supplement CAPTCHA-style proof-of-human checks. According to TechCrunch, the round was led by Chicago Ventures, with participation from BAM Ventures, Aglaé Ventures, Corazon Capital, Tusk Ventures, Yonder Ventures, Rule30 and Brown Alumni Group. Dag Kittlaus, who co-created Siri, is backing the company and will also act as an advisor.

Who is behind it

The company was founded by Alex Oberg, Anna Dorofiyenko and Mike Hall. TechCrunch reports that Oberg previously worked at Ticketmaster, which acquired his earlier startup, and contributed to Verified Fan, the ticketing platform's program for keeping bots out. Dorofiyenko led data science teams at Live Nation and Zillow, while Hall worked at Nike and Tinder, where he helped develop the initial matching algorithm.

Oberg told TechCrunch that as AI advanced over the past two years, he concluded it would become increasingly difficult to separate humans from machines online. In his view, internet identity has always depended on proxies such as email addresses, phone numbers, IP addresses and device fingerprints, and the rise of AI agents means the industry now needs biometric signals like voice instead.

How the technology works

Helix's core model runs directly on a user's device, determining whether a speaker is human and estimating their age. The company says the model can run locally even on hardware as old as an iPhone 13, which keeps deployment costs minimal. The startup is also running pilots to explore whether voice can act as a biomarker for states such as drunkenness or distress, according to TechCrunch.

The technology is already in use at Zeel, an in-home healthcare provider that relies on Helix to confirm the identities of caregivers visiting customers' homes. Verifying workers such as cab drivers is another use case under exploration. Oberg also pointed to age-verification rules for social media platforms, arguing that regulation is effectively compelling users to surrender privacy in exchange for access, and that a voice challenge of about four seconds could offer an alternative.

Acknowledged limitations

Helix is upfront about what its technology cannot do. Oberg told TechCrunch it cannot reliably detect whether someone is lying, it does not identify individual speakers, and he claimed that even if the company stored a voiceprint, it could not be used to reconstruct a person's voice. The company also says the technology is not yet capable of reliably detecting AI agents during live phone calls, a notable gap given the growth of AI-powered voice fraud.

Why Kittlaus is interested

Kittlaus told TechCrunch that voice identification has finally reached the point where it works reliably, positioning Helix to scale as AI-powered fraud increases. He also credited the company's work on compressing its models with giving it a cost advantage. His framing goes beyond ordinary speech recognition: rather than teaching computers to parse words, Helix analyses what a speaker's voice signals about them and their condition. Notably, Kittlaus's description touched on detecting deception, an area the company itself says it cannot yet handle reliably.

Why it matters

The spread of AI agents is eroding the proxy signals that platforms have long used to tell people from bots, and CAPTCHA-style checks are becoming easier for automated systems to defeat. At the same time, photo- and document-based identity verification asks users to hand over sensitive personal information. Helix proposes a third path: an on-device biometric check that, according to the company, does not identify individuals and could satisfy age-verification mandates without an ID upload. Whether voice can hold up against synthetic speech, and whether regulators and users will trust it, remains an open question, and the startup itself admits it cannot yet detect AI agents on live calls.