Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
Businesses IT

AI Startup Revealed To Be 700 Indian Employees Pretending To Be Chatbots (latintimes.com) 55

Posted by BeauHD from the face-palm dept.
An anonymous reader quotes a report from the Latin Times: A once-hyped AI startup backed by Microsoft has filed for bankruptcy after it was revealed that its so-called artificial intelligence was actually hundreds of human workers in India pretending to be chatbots. Builder.ai, a London-based company previously valued at $1.5 billion, marketed its platform as an AI-powered solution that made building apps as simple as ordering pizza. Its virtual assistant, "Natasha," was supposed to generate software using artificial intelligence. In reality, nearly 700 engineers in India were manually coding customer requests behind the scenes, the Times of India reported.

The ruse began to collapse in May when lender Viola Credit seized $37 million from the company's accounts, uncovering that Builder.ai had inflated its 2024 revenue projections by 300%. An audit revealed the company generated just $50 million in revenue, far below the $220 million it claimed to investors. A Wall Street Journal report from 2019 had already questioned Builder.ai's AI claims, and a former executive sued the company that same year for allegedly misleading investors and overstating its technical capabilities. Despite that, the company raised over $445 million from big names including Microsoft and the Qatar Investment Authority. Builder.ai's collapse has triggered a federal investigation in the U.S., with prosecutors in New York requesting financial documents and customer records.
This discussion has been archived. No new comments can be posted.

AI Startup Revealed To Be 700 Indian Employees Pretending To Be Chatbots More

AI Startup Revealed To Be 700 Indian Employees Pretending To Be Chatbots

Comments Filter:

  • A.I. (Score:5, Funny)

    by Local ID10T ( 790134 ) <ID10T.L.USER@gmail.com> on Tuesday June 03, 2025 @04:29PM (#65425397) Homepage

    Actual Indians

  • Wait till the break into OpenAI's HQ... (Score:5, Funny)

    by greytree ( 7124971 ) on Tuesday June 03, 2025 @04:41PM (#65425421)
    ...Full of Mexicans with encyclopedias and not an Nvidia GPU in sight.

    It was fun while it lasted.
  • ...were secretly three raccoons in a trench coat?

  • Even with gen AI. Hilarious. You been techno scammed! What next? Electric bananas we need electric bananas. Those suckers will but anything.

    • Re: (Score:2)

      by tsqr ( 808554 )

      Even with gen AI. Hilarious. You been techno scammed! What next? Electric bananas we need electric bananas. Those suckers will but anything.

      Already done, years ago:

      Electrical banana
      Is gonna be a sudden craze
      Electrical banana
      Is bound to be the very next phase

      -Mellow Yellow, Donovan (1966)

  • Fake it with real intelligence!

    Well, real, wetware brains, anyways. I'm not so sure how intelligent this operation was.

  • Amazon Go (Score:4, Insightful)

    by TomClancy_Jack ( 638962 ) on Tuesday June 03, 2025 @04:51PM (#65425451)
    Reminds me of when Amazon had those Go stores with hundreds of cameras and claimed that AI with computer vision would tally up your cart and keep track of produce etc. But in reality was just people from India watching the video feeds. I wonder how many more AI companies are just Mechanical Turks.

    • Re: (Score:2)

      by Tablizer ( 95088 )

      Do note it wasn't profitable cheating that way, but they hoped enough venture capital would fund R&D to get the real AI good enough. Thus, they were buying time.

    • What do you mean claimed? AI (Actual Intelligent) humans were combined with Computer Vision (camera transmitting mp4 video), to make this work. It was exactly as advertised. :-)

  • I was completely fooled, until I ran out of tokens and it asked me to do the needful.

  • AI...still no use-case at scale.

    Legal, Productive, Profitable. If your business is using AI then you can only pick 2.

    • So the hundreds of millions of people (probably nearly a billion now) using AI on a weekly basis have no use-case? Your brain can believe something like that?

      • Well, nobody's denying the lawyers using it to save time writing briefs that mention court cases that don't exist, and students trying to get one up on their professors, and professors using it to try to find which submissions were written by LLMs, and bad computer programmers using it to write bug ridden security-compromised code without needing to know the language or environment, and websites using it to create websites that contain text to fool search engines into looking like informative content so the

    • Huh?

      This story certainly doesn't prove that.

      Millions of us, like me, use AI every day to make our jobs easier for ourselves, whether that's writing code with GitHub Copilot, or producing slide decks for a business presentation, or getting quick answers to dumb stuff posted by slashdot users.

      I'd say the use cases at scale, are beyond number already.

      • If Microsoft, one of the largest hyper-scalers at the top of the tech industry can't tell the difference between a product a scam, what makes you think you can tell the difference? The smart money telling us that this is a billion dollar industry, are the same guys that can't tell.

        This article is about a Microsoft family for a very basic AI scam. Amazon has also got caught falling for the same thing. And if you keep an eye on indictments and SEC filings, you'll see quite a few similar scams being prosecute

        • So you are asserting that AI is nothing but a scam?

          Here's how I can tell the difference between a real product and a scam: AI products provide real value to me.

          When I use GitHub Copilot, it significantly shortens my coding time. As an example, I've been doing some SQL Server work with XML blobs. The SQL syntax to update or query XML, is arcane and has frustrating limitations. GitHub Copilot is consistently able to generate the proper code, saving me hours of frustration. Once it generates the code, I know e

          • No one believes you. Scam ass marketing comments. Yuck, dude. Chatbots are unproductive. Google had to use their chatbot tools to fill their search engine with degenerate spam so that they can sell you Gemini to cut through it.

            Gemini is like a guy choking out a forest with vines so they can sell more machetes. None of this shit is productive. When any business replaces an employee with a chatbot, they loose customers, revenue and gain reputational harm when those bots start telling people to kill themselve

            • Wow, you are way out there on a limb. Do scams exist? Yes. Is all of AI a scam? Hardly.

              On one point we agree: "When any business replaces an employee with a chatbot..." This idea of replacing employees with chatbots is nothing but marketing hype. AI can't do that, and won't be able to do that for a long time. But what AI *can* do, is increase the productivity of existing employees.

              How do I know that my GitHub Copilot code suggestions aren't a scam? Because there is no way a human could read my code, come up

              • You keep trying to bait me into some argument about your personal experience and I'm not a troll or a rube so I'll never engage on that. AI and crypto are prominent examples of scams. The markets you're trying to pumping are all collapsing. Grow up.

                • I'm not a troll or a rube

                  If you have to say it...

                  While you go around insisting that the earth is flat, the rest of the world will move on. (We do agree on crypto, however.)

                  Personally, I'm thoroughly enjoying this latest technology revolution. It's given me amazing new tools that I didn't believe possible. It's like 1994 all over again.

                  • Gemini is just a search engine that works half as well as Google used to work in 1994. The hallucinations are serious and disqualifying. I was talking to someone about the safety of various cannabis products, including vapes. The industry uses thinning agents in cannabis oils to make them work in disposable vapes a lil better. Personally, I don't believe any of these thinning agents are safe.

                    I asked gemini what's a safe thinning agent for cannabis oil and it suggested MCT oil vaping. Then I asked it if it

                    • I was talking to someone about the safety of various cannabis products,

                      Your responses are starting to make more sense now.

                      Hallucination is a thing with AI. We all know that. But it doesn't nullify the benefit or helpfulness of the technology, it's more a side effect one must be aware of and account for. It's kind of like how actual humans sometimes make things up. We all know this happens, and we acknowledge that it happens and take measures to accommodate this flaw. This flaw doesn't mean that AI, or humans, are useless.

                      Even if you don't find Gemini useful, there are many oth

  • Reminds me of the real mechanical turk [wikipedia.org] (not the Amazon crap), which was a 16th century contraption supposedly using advanced mechanical intelligence to play chess.

    It was really just a guy in a box.

  • Who would use it more than once? (Score:3, Interesting)

    by SuperKendall ( 25149 ) on Tuesday June 03, 2025 @05:40PM (#65425597)

    The summary claimed the company had $50m in revenue (the real number, not corrected).

    I can't understand how it got any revenue, ever - if you ask any real AI to produce code you'll have results in a minute or so.

    But if it was backed by people writing real code, answers would have taken many minutes to hours to produce! Heck just the time to write a summary of the request would seem awfully long.

    Who would use that after any trial? Who was paying them at all?

    Or was it 700 engineers each with a trial chatGPT account just pasting questions and answers back and forth between user and chatGPT?

    • > if you ask any real AI to produce code you'll have results in a minute or so.

      > But if it was backed by people writing real code, answers would have taken many minutes to hours to produce! Heck just the time to write a summary of the request would seem awfully long.

      Maybe the code worked? No functions or language features were hallucinated? No obvious security holes and other bugs were in the code?

  • And yet has no famous large language model. Yet they are building one, a sovereign one no less, which is a good idea for serving poor populations, but if Trump ever finds out about it, watch out for US sanctions for being too socialist.

    Building India’s sovereign model from the ground up is a significant step toward Atmanirbhar Bharat. The sovereign model will be built, deployed, and optimized in India, using local infrastructure and developed by a new generation of Indian talent. This initiative aims

  • It is still artificial intelligence

  • Isn't their business model actually better than all the outsourcing offshore fly by night IT shops AND the half baked output from myriad beta quality AI agents ??

    You are getting your stuff done without delays, near real time i am guessing, and without having to deal with strange accents. And stranger excuses.

    No hallucinations, barring few people on real weed.

    No melting GPUs and 1000s of MW nuclear power plants, just 2000 kcal per unit of staple Indian food. Even if the spices are nearly radioactive.

  • Not AI but (A)I or symply "I", drop the "A" :-)

  • Microsoft wll re-hire these "chatbots" in the wake of their layoffs...

    JoshK.

  • What was the guys expectations?
    He knew it was fake, he knew it was being investigated. Why not transfer millions to various bank in other countries. Transfer it some more, buy a bunch of bitcoins, and take it out as cash. Then get lost. He had access to close to $50 million, even with all transfers and fees for all that he would of been able to do at least $10 million and been able to get to some place relatively safe.

    • Re: (Score:2)

      by MikeS2k ( 589190 )

      Yeah I wonder how many startups are just total scams... Why bother working 8 hours a day, 5 days a week (over 50% of your waking adult life) - when you can start a hilariously dumb startup - e.g. Juicero - even if the thing totally falls over, you've already paid yourself millions over the few years it was in operation, and when it fails you aren't on the hook for that money - you've made more than you can earning honestly. Part of why our society is such a fraud - it does not reward hard work, it reward

  • I'm surprised they weren't Turks.

  • Startup failed because actual chatbots do a better job.

  • Fantastic story! I bet they are not the only ones doing this...

  • If I read this correctly it means....

    The AI was inside us all along!

  • I find this story as funny as the next guy, if the next guy is a guy who finds the story very funny. But the way the story is reported in most places, including here, is exaggerated, bordering on disingenuous.

    If you look at the company's marketing materials, they never claimed AI was going to build your app. Their sales pitch was "human-assisted AI". You were supposed to work out your solution with a chatbot, after which a human dev was going to take over and develop it for you. Supposedly more cheaply beca

Slashdot Top Deals

Drilling for oil is boring.

Close