Helix Raises $6.2M for Voice-Based Proof of Humanity
Helix wants to replace the CAPTCHA with four seconds of speech. The startup has raised $6.2 million to sell voice technology that checks whether a user is human without collecting sensitive personal data.
The round and the team
Chicago Ventures led the round. BAM Ventures, Aglaé Ventures, Corazon Capital, Press On Ventures, Tusk Ventures, Yonder Ventures, Rule30 and Brown Angel Group also took part. Siri co-founder Dag Kittlaus is backing the company and will serve as an advisor.
Alex Oberg, Anna Dorofiyenko and Mike Hall founded Helix in 2025. Oberg previously worked at Ticketmaster, which acquired his startup, and helped create Verified Fan, a tool designed to keep bots off the platform. Dorofiyenko led data science teams at Live Nation and Zillow. Hall has worked at Nike and Tinder.
What the technology does
The core model runs on the user's own device. It decides whether a person is human and estimates their age. Because it can run locally even on an iPhone 13, Helix says the cost of deploying it is minimal.
Helix says its voice check can stand in for CAPTCHA or work next to it, and can help stop bot attacks. One customer is Zeel, an in-home healthcare provider that uses it to verify providers visiting customers' homes. Oberg said the startup is also exploring checks on workers, cab drivers among them.
Helix is also running pilots to test whether voice can serve as a biomarker for conditions such as drunkenness or distress.
The privacy argument
Oberg's pitch starts from a change in how identity works online. "Identity on the Internet's always been about proxies for identity, right?" he said, listing email, phone number, IP address and device fingerprint. As AI agents spread, he argues, the industry needs signals like voice.
He also tied the product to age-verification rules for social media around the world, which he said force users to trade privacy for access. In his words, Helix is the company saying no to that trade.
The limits are stated openly. Oberg said the technology cannot dependably catch a lie. It does not identify individual speakers, he said, and a stored voiceprint cannot be used to rebuild a person's voice. The company also believes it cannot yet reliably spot AI agents during live phone calls, even though AI-powered fraud on calls is a growing problem.
Kittlaus said voice identification has finally reached the point where it works reliably, and that Helix's work compressing its models gives it a cost advantage.
Source: TechCrunch