Fintech founder charged with fraud; AI app found to be humans in the Philippines

Original Article ↗ Hacker News Discussion ↗

What crossed the line into fraud

  • Commenters emphasize the key issue: not using humans, but lying about it to investors.
  • DOJ materials referenced in the thread say the founder claimed 93–97% automation “without human intervention” when internal reality was “effectively 0%” automation.
  • Access to an “automation rate dashboard” was allegedly restricted and framed as a “trade secret,” reinforcing the idea of deliberate concealment.
  • Several people note that human fallbacks are normal (Waymo teleassist, Amazon Go reviewers, RLHF labelers); fraud begins when you pitch “edge cases” but in fact everything is an edge case.

Humans behind ‘AI’ as a general pattern

  • Many examples raised: Amazon’s Just Walk Out tech using hundreds/thousands of reviewers, mechanical turk workers, click farms, offshore video reviewers, and “AI” customer operations that are really BPOs in disguise.
  • Running joke acronyms: “AAI: Artificial Artificial Intelligence,” “AI = Actually Indians / All Indians / A Guy Instead.”
  • Some argue MTurk itself is at risk: once LLM-using workers are indistinguishable from honest workers, quality control collapses and the economics may no longer work.

Startup dynamics: from ‘do things that don’t scale’ to deception

  • Multiple commenters outline a recurring trajectory:
    • Prototype ML works for a narrow case → launch startup.
    • It fails to generalize → humans fill in “edges” to preserve reputation.
    • Human pipeline quietly becomes the core system → temptation to keep claiming AI and raise more money.
  • In this case, people stress it went further: claims of sophisticated models (LSTM, NLP, RL) and high automation, with essentially no working model behind it.

Investor behavior and due diligence

  • Many are baffled that tens of millions were invested without verifying automation rates or demanding real access to metrics.
  • A common view: investors overweight charisma, elite credentials, and hype (“fintech,” “AI”) over technical diligence.
  • Some note a cynical asymmetry: regulators act when wealthy investors are harmed, while consumer deception and big-company hype (Amazon Go, Tesla FSD, adtech “AI”) rarely face similar consequences.

Broader reflections on AI hype and feasibility

  • Commenters see this as part of a wider AI boom pattern, comparable to crypto: glossy promises, weak tech, and marketing-first founders.
  • Several argue many “AI will automate X” startups are structurally doomed because counterparties actively resist being automated and keep changing flows to break bots, forcing humans back into the loop.