Fake AI startups? Founder charged for fraudulent AI app powered by human labour

A former tech darling once hailed for its promise of seamless e-commerce has become the latest cautionary tale in the high-stakes world of artificial intelligence.

Albert Saniger, founder and former CEO of Nate, has been charged by the US Department of Justice for misleading investors about the capabilities of his company’s AI-powered shopping app.

The app, which claimed to enable users to complete purchases from any online store with a single click, allegedly relied on human labour rather than cutting-edge AI.

The DOJ’s indictment, released this week, paints a picture of smoke and mirrors. Nate raised more than US$50 million in venture funding, including a $38 million Series A round led by Renegade Partners in 2021, with other backers such as Coatue and Forerunner Ventures.

But rather than a sophisticated AI engine behind the scenes, the app reportedly leaned heavily on manual processing, with hundreds of contractors in the Philippines executing customer purchases.

While Nate presented itself as a shining example of AI innovation – promising a checkout experience “without human intervention” – the DOJ alleges that its actual automation rate was effectively 0%.

The only technology in play was purportedly human intelligence, operating out of offshore call centres. Investors were led to believe the AI was handling transactions seamlessly, when in reality, people were pulling the strings behind the curtain.

Despite hiring data scientists and acquiring some AI technology, the app’s automation never took off. Nate eventually ran out of funds and sold its remaining assets in January 2023, leaving investors with what the DOJ described as “near total” losses.

Saniger is no longer listed as the company’s CEO as of 2023, according to his LinkedIn profile. Currently a managing partner at New York-based VC firm Buttercore Partners, Saniger has not responded to requests for comment.

Read: CEO wants proof you're better than a bot

A pattern of AI puffery

Nate’s story is far from unique. In recent years, a string of startups have been accused of overstating their AI prowess, relying on human labour while riding the AI hype wave to secure funding and media attention.

What seems like magic on the surface is often just elbow grease in disguise.

In 2023, The Verge uncovered that a drive-through AI software startup – ostensibly offering machine-powered voice ordering – was also run largely by humans based in the Philippines. Similarly, legal tech unicorn EvenUp was reported by Business Insider to be relying on human input for much of its so-called AI-driven work.

Rewind to 2019, and the Wall Street Journal reported that Indian startup Engineer.ai had been promoting itself as an AI-enabled app builder while primarily using human engineers. Backed by nearly $30 million, including funding from a SoftBank-owned entity, the startup claimed its tools could build over 80% of a mobile app from scratch in under an hour.

Sachin Dev Duggal, Engineer.ai’s founder and self-styled “Chief Wizard,” positioned the product as “human-assisted AI”. However, insiders claimed the platform had barely begun to develop meaningful automation. The company’s former chief business officer, Robert Holdheim, even filed a lawsuit alleging the firm was misrepresenting its capabilities to attract funding.

When pressed on the nature of its technology, Engineer.ai admitted to using basic machine learning methods – natural language processing to estimate project costs and timelines, as well as decision trees to allocate tasks. These features fall far short of what most industry experts would classify as genuine AI. The software, it turns out, didn’t generate code; that was still the job of engineers in India and other regions. The company did not respond to further inquiries.

Human-powered ‘AI’ and the startup shortcut

The broader pattern suggests an industry increasingly plagued by “pseudo-AIs” – startups using human effort to simulate AI capabilities while banking on future breakthroughs.

As Gregory Koberger, CEO of developer tool firm ReadMe, observed:

Using a human to do the job lets you skip over a load of technical and business development challenges.

“It doesn’t scale, obviously, but it allows you to build something and skip the hard part early on,” Koberger said, as quoted by The Guardian.

This workaround – using people to impersonate machines – isn’t new. Back in 2016, Bloomberg reported on calendar scheduling startups where workers spent 12-hour shifts pretending to be AI chatbots.

The work was allegedly so monotonous that some employees longed for actual AI to put them out of a job. In 2017, expense management app Expensify also admitted that its “smartscan” tech was being supplemented by humans transcribing receipts via Amazon’s Mechanical Turk, a crowdsourced platform known for low-wage digital piecework.

These revelations speak to a broader issue in the tech ecosystem: building real AI is both capital-intensive and technically complex. Faced with mounting pressure to scale and satisfy investors, some founders take the path of least resistance – substituting hype for horsepower.

Read: Is Gen AI really making you productive?

AI investment boom fuels exaggerations

The temptation to overstate AI capabilities comes at a time when the sector is flush with investor interest. According to PitchBook, funding for AI startups reached $31 billion last year. Meanwhile, major players like SoftBank have pledged to pour hundreds of billions into AI innovation over the coming years.

Startups have responded by branding themselves with the .ai domain and wrapping traditional tech offerings in AI terminology. In a landscape dominated by behemoths such as Google, Meta, and Amazon, claiming to be on the bleeding edge of AI offers a way for smaller players to capture investor attention and consumer interest – at least temporarily.

But the fallout from these exaggerated claims may be starting to catch up. As more cases come to light, regulators and investors alike are scrutinising claims more closely, and the line between innovation and misrepresentation is becoming harder to blur.

The rise and fall of Nate, along with similar stories from across the tech sector, suggests that what’s marketed as a leap into the future may just be a clever sleight of hand. In the rush to capitalise on the AI gold rush, some founders are skipping the real AI development work, swapping ambition for artifice – and leaving investors to pick up the pieces.