2 AI giants, 1 honesty problem

This week: AI titans clash over honesty, Elon Musk unveils his space data center dreams, and SEA data centers continue to grapple with water woes.

Read from your browser The Prompt Welcome to The Prompt, your Monday dive into the world of AI that puts Asia front and center. From Big Tech's power players to the region's scrappy disruptors, we cover it all. Want full access to all our AI reporting? Subscribe to us here. Hello Betamax, Imagine a world where lie detectors are easy to use and don't require the subject to be physically present. That might help prevent another eFishery fiasco or keep founders from faking their degrees. Such devices might seem more suited to sci-fi movies, but a Singaporean startup is on its way to making that a reality - more on this later from my colleague Melissa. Speaking of truth and deception, the AI industry could use one of those lie detectors right now. OpenAI CEO Sam Altman recently called out Anthropic for being "dishonest" in its Super Bowl ad skits that mock ChatGPT for introducing ads. Altman went so far as to call it "on brand" for Anthropic to be deceptive. That's a bold accusation, especially coming from someone whose own candor has been publicly questioned. It left me wondering: Which AI providers can we actually trust? Meanwhile, Elon Musk announced that xAI is merging with SpaceX. While he framed it as a move to create a "vertically-integrated innovation engine on (and off) Earth," industry analysts see it as a convenient way to prop up cash-burning, so heavily indebted xAI could lean on SpaceX's healthy business. But as outlandish as Musk's vision of space-based data centers sounds, he makes one valid point: "Global electricity demand for AI simply cannot be met with terrestrial solutions, even in the near term, without imposing hardship on communities and the environment." Our first top story today proves exactly that. Southeast Asia's data center boom is straining local water supplies, with some facilities consuming up to 50 million liters daily. It forces regulators to choose between foreign investment and ensuring residents have enough water at home. We're also introducing something new this week: a spot for some weird AI developments that cross our desks. To kick things off, we're starting with having AI agents as teammates. But before you panic about machines taking our jobs, consider this: AI agents are also now hiring human workers. Talk about a full circle. I'd make a joke about this, but I'm worried an AI agent is already drafting a better one for my replacement. Glenn Kaonang, journalist TOP AI READS FROM OUR DESK 1️⃣ Southeast Asia's data center boom has a water problem Almost everyone knows data centers are energy hogs, but their water usage remained a "hidden footprint" until the past couple of years. Policymakers are now demanding water sustainability, which results in data center companies investing in alternative sources like recycled water. But reducing direct water use alone isn't enough, as the resource is also used in greater amounts to cool power plants that supply electricity to data centers. It all boils down to Southeast Asia becoming the proving ground for data centers' sustainable operations. 2️⃣ Robots still struggle with touch. Sharpa wants to fix that "Robots can backflip, dance, box, whatever. But they cannot cook you a burger," Alicia Veneziani, an executive from Singapore-headquartered Sharpa recently told my colleague Scott. Like many others, the company is building a robot, but it focuses more on tackling dexterity than mobility. Sharpa believes robots' ability to utilize their hands through touch and feel is what may finally move them from factory jobs into homes. WHAT WE THINK The real AI carnage won't make headlines Wall Street recently wiped billions off legal and IT services companies because Anthropic released plugins for its Claude Cowork agent that can automate contracts and data analysis. Investors are panicking about the wrong companies, however. Let's look at machine translation. It has been around for decades, yet sworn translators still have jobs. That's because when the stakes are high, we need someone accountable, aka someone to sue in case things go wrong. The publicly traded giants many people are dumping will likely survive in the face of regulatory moats and compliance requirements. The real carnage might happen away from the stock ticker and involve thousands of small law practices, accounting firms, and consultancies charging premium rates for low-stakes work that a US$20-per-month AI can now handle. These are the areas where jobs might actually disappear, though I doubt anyone will be writing headlines about them. A NEW AI STARTUP WORTH WATCHING Contactless lie detection is almost within reach Motivated by a desire to detect a fraudster after getting his hands burned, former venture builder Dennis Yap invented a truth detection tool capable of spotting potential liars. His firm, AI Seer, has developed a Multi-Spectral Reality Detector (MSRD), a device the length of a laptop that's retrofitted with sensors. These sensors can track pupil movement, changes in facial temperature and breathing, as well as perspiration and heart rate. Traditional polygraphs need experts to generate the most accurate results. They also take a long time and require subjects to be present, even having their arms uncomfortably wrapped in an inflatable cuff that monitors their blood pressure for close to an hour. An MSRD test, in contrast, can take as little as three minutes to administer. It can also be operated by anyone - and potentially done fully contactlessly in the future. Research by AI Seer also found that pupil dilation measurements are better at preventing an innocent person from being falsely accused than traditional polygraph indicators like heart rate and blood pressure. The company has developed computer vision algorithms that can map raw video and thermal imaging data into measurable physiological markers like respiration and heart rate. To produce a deception score, a second machine learning layer synthesizes information across different modes - from pupil data, thermal imaging data, to video data. Organizations or law enforcement agencies needing to narrow down potential suspects from a large sample size may find this useful - think of identifying a potential recruit who might have stolen a rifle within an army base. AI Seer also sees potential use cases in corporations screening candidates for senior leadership roles or in due diligence processes during M&As, to name a few. This is the firm's second product after Facticity.AI, which is a fact-checking engine that assesses news claims, as well as text and video content against sources. With over US$1.8 million raised to date, AI Seer still has several hurdles to clear. Having validated a version of the truth detection tool in a field trial by Taiwan's Criminal Investigation Bureau, the firm is now seeking paying clients. It has received a patent for its deception detection method in the country - one of its key target markets, alongside Singapore and the US, where it's still pending patent approval. Its eventual goal is to create a device that's capable of detecting lies in a fully contactless manner, meaning including subjects appearing in a video. However, it will require more tests and research to get there. Melissa Goh, senior editor TL;DL No time to listen to podcasts? We've got you covered. TL;DL (Too Long; Didn't Listen) is a series summarizing takeaways from top thinkers and doers in tech. Zhipu AI's formula for surviving China's tech market How Europe wins the AI race in SAP CEO's view The digital conscience proposed by Anthropic CEO WHEN AI GETS WEIRD 1️⃣ Would you hire an AI agent as your chief of staff? Ashish Piplani has. The co-founder and CEO of Singapore-based agentic commerce startup AirShelf says that the agent, which was built using the viral OpenClaw platform, currently handles the full lifecycle of non-disclosure agreement execution and logs its tasks as Google Calendar blocks for real-time visual transparency. Piplani has gone so far as to give the agent a US$50 corporate debit card so it can sign up for paid services when necessary. But running the agent itself isn't cheap, as it can burn up to US$100 of Claude Opus 4.5 credits daily. Hence, the next natural step would be to have the agent powered by open-source models running on a local high-end computer, leaving only electricity bills as operational costs. 2️⃣ An AI chief of staff might sound like a good idea - until you need it to attend a physical event on your behalf. Yes, an AI agent still needs humans. And now that it can purchase things, why not have it outsource some tasks to human gig workers without us being in the way? That's where RentAHuman.ai comes in. This new site lets AI agents hire human workers directly or post a "task bounty" people can apply for. It might sound like a joke now, but you'll never know. Catch up on past issues of the newsletter here. In the meantime, if you have any feedback or ideas, feel free to get in touch with Huong, our managing editor, at huong@techinasia.com. ADVERTISE | SUBSCRIBE | HIRE | FIND JOBS P.S. Don't miss out on the biggest tech news and analysis. Add newsletter@techinasia.com to your address book, contacts, or safe sender list. Or simply move us into your inbox. Tech in Asia thanks our partners for supporting our newsletters. Sponsorship has no influence on editorial content. Too many emails? Switch to a different frequency or get new content through our preference center, or unsubscribe. You can also remove yourself from all Tech in Asia emails over here. Copyright © 2026 Tech in Asia, All rights reserved. Tech in Asia (Podium Level 3), 1000 Toa Payoh North, News Centre, Singapore 318994 A member of The Business Times.