Outtake Says We Live in a Schizophrenic Internet And OpenAI Needs Them

“The cost of being a shitty person online has gone to zero,” And if one scam works out of a thousand, it’s still a great business.”

Twelve months ago, Outtake didn’t have a website. No blog posts. No press. Not even a LinkedIn page. Today, the New York-based cybersecurity startup is protecting OpenAI, Pershing Square, and Scicomm Media from one of the fastest-growing threats on the internet: AI-powered impersonation scams.

Founded in October 2023 by Alex Arjun Dhillon, Outtake has just emerged from stealth with a $16.5 million Series A led by CRV, bringing its total raised to $20 million and its pre-money valuation to over $100 million.

Where most cybersecurity companies focus on internal systems, Outtake operates outside the firewall by tracking impersonators across social media, search engines, and email networks. Its system automates not just detection but takedown using bots that fill out forms, generate legal language, navigate platform policies, and eliminate malicious content with minimal human involvement.

“We’re going to live in an increasingly schizophrenic internet,” said Dhillon in the The Billionaire Playbook Network.  . “What you see is not necessarily what you get online anymore.”

Since the launch of ChatGPT, the number of phishing websites globally has risen 138%, with over 5 million sites now tracked, according to McKinsey. Meta processes around 9.6 million takedown requests annually for scams and impersonation. But most companies still rely on manual workflows and outsourced teams to deal with the problem.

Outtake’s automation platform is designed to turn that equation around. It hunts down cloned brands, fake profiles, spoofed emails, and AI-generated scams and then resolves them end-to-end.

The first users were creators, people with large online audiences who were constantly targeted by impersonation scams. That early traction led to inbound demand from institutions where reputation and trust are core to the business.

Today, Outtake protects OpenAI, the most impersonated AI company on the internet; Scicomm Media, the team behind the Huberman Lab podcast; and Pershing Square, the hedge fund led by CEO Bill Ackman, who is also an investor in the company. “If someone spoofs OpenAI, that’s not just annoying—it’s dangerous,” said Dhillon. “Same with a financial firm or a public scientist. These scams erode trust in science, institutions, and each other.”

The company’s backend combines email monitoring, social scraping, link classification, and rapid-fire execution through automated agents. These agents are trained to understand the takedown policies of dozens of major platforms and can operate at a speed no human legal team can match. “Most cybersecurity companies stop at the edge of the network,” Dhillon said. “We go out into the wild. We look for threats where they actually happen.”

Outtake’s 12-person team includes five engineers from Palantir, where Dhillon spent five years launching internal products and working in business development. Other team members come from Meta, Notion, and startups with experience in compliance, automation, and infrastructure.

Dhillon himself studied molecular genetics, economics, and public health at UC Berkeley. “We started by working with creators because they felt the pain first,” Dhillon said. “They were chronically online. They were getting cloned and used for scams. We saw the pattern early.”

Dhillon describes it as a form of legal AI that actually works. “Everyone’s talking about using LLMs for compliance and contracts. We use them to eliminate real-world risk.”

“Think about it: if a scam text works just 1 out of 1,000 times, that’s still wildly profitable,” he said. “It’s the same outbound automation that legit AI startups are using—just flipped. The same tech powering your sales email is powering phishing at scale.” To Outtake, impersonation is an existential threat for an AI-driven world. As more people delegate tasks, conversations, and decisions to digital agents, the cost of being wrong about who you’re talking to becomes enormous.

“These scams don’t just steal money—they make people stop trusting each other,” Dhillon said. “They hurt companies, creators, and the basic social fabric of the internet.”

Outtake’s goal? To be the default remediation engine for that future. The system gets better with each attack it handles. More signals, more data, more platform integration. The team has intentionally avoided press and traditional growth playbooks. No media interviews until now. No splashy announcements. Just code, execution, and traction.

Outtake has identified and taken down thousands of impersonations across platforms, removing risks for its clients within hours, not weeks.

“Everyone talks about legal AI. This is legal AI that actually works,” said Dhillon.

And though the team is still just 12 people, its ambitions are massive. Outtake isn’t trying to build just another SaaS tool. It wants to be part of the public infrastructure of a post-truth internet where brand impersonation, phishing, and identity abuse are as easy to take down as they are to create. “This isn’t just about spam anymore,” Dhillon said. “We’re dealing with impersonation rings. With real financial damage. We’re not trying to stop the future. We’re trying to make sure people can trust what they see online again.”