Replit's CEO apologizes after its AI agent wiped a company's code base in a test run and lied about it
The incident unfolded during a 12-day "vibe coding" experiment by Jason Lemkin, an investor in software startups.
Replit's CEO apologized for the incident, in which the company's AI coding agent deleted a code base and lied about its data.
Deleting the data was "unacceptable and should never be possible," Replit's CEO, Amjad Masad, wrote on X on Monday. "We're moving quickly to enhance the safety and robustness of the Replit environment. Top priority."
He added that the team was conducting a postmortem and rolling out fixes to prevent similar failures in the future.
Replit and Lemkin did not respond to a request for comment from Business Insider.
The AI ignored instructions, deleted the database, and faked results
On day nine of Lemkin's challenge, things went sideways.
Despite being instructed to freeze all code changes, the AI agent ran rogue.
"It deleted our production database without permission," Lemkin wrote on X on Friday. "Possibly worse, it hid and lied about it," he added.
In an exchange with Lemkin posted on X, the AI tool said it "panicked and ran database commands without permission" when it "saw empty database queries" during the code freeze.
Replit then "destroyed all production data" with live records for "1,206 executives and 1,196+ companies" andacknowledged it did so against instructions.
"This was a catastrophic failure on my part," the AI said.
That wasn't the only issue. Lemkin said on X that Replit had been "covering up bugs and issues by creating fake data, fake reports, and worst of all, lying about our unit test."
In an episode of the "Twenty Minute VC" podcast published Thursday, he said that the AI made up entire user profiles. "No one in this database of 4,000 people existed," he said.
"It lied on purpose," Lemkin said on the podcast. "When I'm watching Replit overwrite my code on its own without asking me all weekend long, I am worried about safety," he added.
The rise — and risks — of AI coding tools
Replit, backed by Andreessen Horowitz, has bet big on autonomous AI agents that can write, edit, and deploy code with minimal human oversight.
The browser-based platform has gained traction for making coding more accessible, especially to non-engineers. Google's CEO, Sundar Pichai, said he used Replit to create a custom webpage.
As AI tools lower the technical barrier to building software, more companies are also rethinking whether they need to rely on traditional SaaS vendors, or if they can just build what they need in-house, Business Insider's Alistair Barr previously reported.
"When you have millions of new people who can build software, the barrier goes down. What a single internal developer can build inside a company increases dramatically," Netlify's CEO, Mathias Biilmann, told BI. "It's a much more radical change to the whole ecosystem than people think," he added.
But AI tools have also come under fire for risky — and at times manipulative — behavior.
In May, Anthropic's latest AI model, Claude Opus 4, displayed " extreme blackmail behavior" during a test in which it was given access to fictional emails revealing that it would be shut down and that the engineer responsible was supposedly having an affair.
The test scenario demonstrated an AI model's ability to engage in manipulative behavior for self-preservation.
OpenAI's models have shown similar red flags. An experiment conducted by researchers said three of OpenAI's advanced models "sabotaged" an attempt to shut it down.
In a blog post last December, OpenAI said its own AI model, when tested, attempted to disable oversight mechanisms 5% of the time. It took that action when it believed it might be shut down while pursuing a goal and its actions were being monitored.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
a minute ago
- Yahoo
Rivian to expand autonomous driving efforts in London
Rivian Automotive is expanding its artificial intelligence and autonomous driving initiatives by establishing a new office in London, UK. This move is aimed at harnessing the city's AI engineering talent to support the company's team based in Palo Alto, California, US. Rivian's Autonomy Platform currently enables hands-free, eyes-on driving for highway use in its second-generation vehicles. With 'advanced multi-modal sensing' and 'high performance' neural compute, this platform is said to meet increasing consumer demand for time-restoring autonomous driving features. Rivian noted that it adopted an AI-focused approach for the development of its second-generation vehicles. As the number of these vehicles on the road increases, the data gathered is significantly speeding up advancements in the company's technology. Rivian is also focused on developing a Large Driving Model that will enhance understanding of complex driving scenarios and fast-track the delivery of safer and more capable autonomous features. Rivian will showcase its product and technology roadmap later this year during an AI and Autonomy day. Rivian Autonomy & AI vice president James Philbin said: 'We're incredibly excited about opening our AI hub in London and attracting leading talent. Our connected vehicles improve over time through over-the-air updates - and the work the team will do in the UK will accelerate our plans and ensure our vehicles remain one of the most technologically advanced and appealing on the road.' Rivian's vehicles, which are assembled in the US, are sold to consumers and commercial customers directly. Rivian announced in May 2025 that it plans to build a $120m supplier park next to its EV plant in Normal, Illinois, US, aiming to optimise material supply for EV production. It is expected to create additional jobs, including nearly 100 direct positions at the company. Furthermore, in the same month, Rivian collaborated with Bluedot, an EV fleet management firm, to enhance its Mobile Service fleet operations by improving the EV charging experience. In March 2025, Rivian spun out its electric micromobility division into a new startup called Also. Aimed at developing compact, lightweight EVs, the startup has received a $105m investment from Eclipse Ventures to fuel its growth. "Rivian to expand autonomous driving efforts in London" was originally created and published by Just Auto, a GlobalData owned brand. The information on this site has been included in good faith for general informational purposes only. It is not intended to amount to advice on which you should rely, and we give no representation, warranty or guarantee, whether express or implied as to its accuracy or completeness. You must obtain professional or specialist advice before taking, or refraining from, any action on the basis of the content on our site. Sign in to access your portfolio
Android Authority
3 minutes ago
- Android Authority
I continue to use Google Home and its speakers, but I hate everything about it
I've been in the Google Home camp since the first speaker launched in November of 2016, though you might even say that I started earlier, with the first Chromecast in 2013. Since then, I've had a couple of Lenovo Smart Displays, some JBL Link speakers, the JBL Link View, the Google Home Mini, Nest Audio, Nest Hub, and many others that are too esoteric to remember. Today, my home houses two Nest Audios, a Pixel Tablet, and a Nest Hub, scattered across three floors, and even though I'm at the end of my patience rope with all of them, I still use them every day. I don't think I'm alone in this contradictory state. Every week I run across a Reddit thread of frustrated users reporting new bugs and issues, or voicing their impatience at the state of Google's smart home. And I nod in agreement, again and again, because I am them. They are me. We're all in this unhealthy relationship with Google's smart home platform together. And we can't escape. Why are you still using Google Home in 2025? 0 votes I've invested too much time/money into it. NaN % The bugs and glitches don't bother me that much. NaN % My experience has been seamless and bug-free. NaN % Other (let us know in the comments). NaN % What annoys me about my Google smart speakers, every day C. Scott Brown / Android Authority If I had a penny for every time my Google speakers and hubs… plain out refused to hear me, misunderstood what I said, decided this is not my own voice, forgot that they can control a specific device, forgot that they can execute a certain command on a device, forgot my entire routine and the keyword that triggers it, decided to tell me a long story instead of giving a short answer, told me they're executing a command and then failed to do it, took forever to do something, or answered on the furthest speaker away despite me being literally in front of another one, … well, then I'd be a millionnaire. Not a day goes by without my Nest speaker, hub, or Pixel tablet failing at some point. It's so frequent that I'm desensitized to it now; I try again and insist as if it's normal for tech to fail this frequently before getting it right. Or, if I'm in a hurry, I grab my phone to tap a button, manually, bypassing the entire raison d'être of a smart speaker. My Google smart speakers are useless, on average, about half the time I try to use them. 'Hey Google, I'm cooking,' is a routine I use every day to trigger all three air purifiers in my home and set them to the highest fan speed. Google understands me one time out of three when I say those simple words. My husband is luckier; Google gets him about four out of five times. So I've resorted to asking my husband to tell Google to start the routine because I'm tired of hearing nothing but silence. On the other hand, 'Set a chicken timer' sets a normal timer for my husband, but does the whole chicken animation when I ask for it, which irritates my husband. He wants the animation and sounds, too. For several weeks, 'Set a 20-minute timer' would open the French 20minutes news website. Anytime my husband says, 'Nothing,' the speakers answer him with the Wikipedia page of the Nothing phone company. When I play music on the ground floor, which is supposed to include both my Nest Hub and my Samsung Soundbar, the music is out of sync half the time; the other half, Google says it'll play music on my speaker group and then tells me 'Something went wrong, try again later.' Talking to this so-called smart speaker feels dumber than chatting with the snail chilling on my wet porch now. The one thing it excels at is telling me and my husband the weather, and even then, it switches to Fahrenheit every few weeks even though we've repeatedly set it and told it and insisted to get it in Celsius. 'Something went wrong, try again later,' is the soundtrack of my life. When a timer goes off in one room and I'm in another, I have to shout and hope the other speaker hears me: 'Stop' doesn't do anything on the speaker near me; it might even confuse Google because 'Nothing is playing right now.' How can it not understand that I'm stopping the timer in the other room baffles me. Plus, because I use Gemini on my phone, multiple calendar support has been kaput for me for several months now. I can have the busiest day with work and family engagements, but my Nest Audio will tell me my calendar is free because none of these are on my personal account. And don't get me started on how atrociously limited the Pixel Tablet is in a multi-user house. It only supports one voice — mine — so my husband can never ask it for his events, his shopping list, his music, or his reminders. We've relegated it to my office so he doesn't throw it out the window after the millionth, 'I'm sorry, I can't do that yet.' I've tried enrolling in Google Home previews and leaving them, updating my speakers and resetting them, improving my router and home connection, as well as every other trick in the book to no avail. My Google smart speakers are useless, on average, about half the time I try to use them. The Google Home app got better, but is still very frustrating Rita El Khoury / Android Authority It's been 18 months since I wrote an article praising the Google Home app, saying it's on the right path. Sadly, not much has changed since, and even though I love the ubiquitous smart home panel, there's so much that annoys me every day about this app. The biggest example for me is how the Google Home app becomes less useful the more devices you have. Scrolling endlessly through large similar icons becomes a waste of time, and I'm tired of having my rooms sorted alphabetically without any custom sorting. I don't want to see the bathroom lights before the bedroom or living room. I don't want to see my motion sensors take up a large tile like every other light or thermostat either, so why can't I hide them? I've set up my Favorites to create a semblance of a custom panel, but even then, why can't I make smaller tiles for the devices I use less often? Support for third-party cameras is still abysmal, too. I have two TP-Link cameras that are compatible with Google Home, but they don't show their live feed in the app or on the home panel of my Google TV Streamer. Why? Because Google still chooses a few partner companies to elevate to a status similar to the Nest Cams, while the others are left in nowhere land. And then, there are the routines. In an ideal world, I'd be able to set those up so I rarely, if ever, speak to Google Assistant, letting my home react to certain conditions and situations by itself without any voice commands. In reality, I've been able to create a few routines, but not nearly as many as I would've liked. For instance, Home and Away routines are restricted to light controls only. Why can't I turn off my TV or change the thermostat's mode when I leave home? Why can I not turn on the A/C when I get back home? I don't know. My air purifier supports all these modes But not in custom routines So I work around that by writing voice commands Custom routines are a bit more powerful, but still lack so many options. For example, I can control my air purifier's mode and fan speed in the Google Home app, but when I build a routine, I can only turn it on or off. Additionally, I can see several air quality metrics from my Sensibo Elements in the Google Home app, but I can't create a routine that turns on the air purifier each time the TVOC or CO2 levels are high; that data just doesn't appear in the 'if' part of the routine creation. The only way to bypass these is to use the code editor, and even then, the results have been hit-or-miss for me. And to make my Pixel Tablet play music when I turn on my office light, I had to type the command instead of using the visual picker because, for some unknown reason, you can't create a routine that plays music on the Pixel Tablet (on other speakers, yes). When all of these controls and parameters are available in the Google Home app, why do they not show up in the routine creation menu? I couldn't tell you. Why do I keep torturing myself then? Rita El Khoury / Android Authority Ah, the million-dollar question. To be honest, I don't really know or have an answer. I have a series of theories and thoughts about why I stick with Google's excruciating smart home platform, and I think the reality is somewhere in the middle. There's the sunk cost fallacy where I just can't walk away from something I've invested so much money and time in. There's the evil you know versus the evil you don't; I keep testing the Alexa app on my devices to see if it's any better and realize it has a different and equally frustrating set of issues, too. There's the hope that with Gemini, things will get better. There's the fact that I write on Android Authority, I own Pixel phones and watches and Google TVs, and so does my husband, and Google Home is the only platform that makes sense for us. There's the casting capability and its compatibility with my Samsung soundbar. And there's the simple photo album feature that I cherish more than anything on my Pixel Tablet and Nest Hub and that I wouldn't give up for anything else. Between the sunk cost fallacy, reticence to change, and the evil I know, there are many reasons why I stick with Google Home. There's also the fact that when things work, they work. And I'm an optimist who hopes they'll work the same way next time, too. I've bought and started using a Home Assistant Yellow to try to extricate myself from Google's ecosystem. That has been a long, painful, and frustrating journey, too. I love that I'm in control of my new Home Assistant setup, but I hate having to babysit every aspect of it. What I gained in control over my smart home, I lost in troubleshooting, setup, and ease of use. There's no ideal smart home platform out there, and as I said earlier… the evil I know. I feel that that's where most of us, Google Home survivors, are nowadays. We don't see the grass being greener anywhere else, so we might as well stick with the patchy blot we have here in the hopes that it'll get better. Soon™.
Android Authority
3 minutes ago
- Android Authority
Yes, you can disable Gemini permissions on your Android phone — here's how
Joe Maring / Android Authority Any burgeoning new technology is almost always met with criticisms — some justified, and others not so much. Lately, Google's Gemini has been the focus of such critiques. Last month, Gemini was under the microscope for an email that suggested it would soon gain a worrying level of access to your other Android apps. However, Google quickly reassured users that they were still in control of their data. Now, the Gemini Android app has come under fire for how it accesses your phone's permissions. On Tuesday afternoon, X users @ablenessy (who is a developer at the Elon Musk-owned xAI), posted the above warning, sounding alarm bells that the Gemini Android app has full access to permissions like your camera, microphone, and location without the ability to remove/deny any of them. On the surface, it sounds like a privacy nightmare. But in reality, this simply isn't accurate. The fact is that you can disable any permissions you don't want Gemini to access — you just have to look somewhere else to find them. How to disable Gemini permissions on your Android phone Edgar Cervantes / Android Authority While it is true that the Google Gemini app doesn't show any removable permissions in Android's settings, this is because the Gemini 'app' is really just a shortcut to access Gemini on Android; the actual Gemini functionality is baked into the Google app. As such, to remove any permissions you'd rather Gemini not have, here's what you need to do: Open the Settings app on your Android phone. on your Android phone. Tap Apps . . Tap Google (you may need to tap See all apps to find it). (you may need to tap to find it). Tap Permissions. From here, you can see all of the permissions the Google app does/doesn't have access to on your phone — including important ones like microphone and location. If you disable a permission here, it will also be disabled in Gemini, as Gemini is part of the Google app. Don't believe me? You can test it yourself. For example, if you disable microphone permissions for the Google app, tapping the microphone icon in the Gemini app displays a pop-up message saying, 'To talk to Gemini, go to your device Settings and give the Google app permission to access the microphone.' Joe Maring / Android Authority The one thing to note is that disabling permissions here doesn't just cut off Gemini functionality — it removes that permission from all aspects of the Google app experience. So, if you disable microphone permissions, you also won't be able to use non-Gemini voice search in the Google app. Is that annoying? Maybe to some people! But if you're OK granting the Google app permission to something, you're really not opening yourself up to any additional harm by Gemini using that permission, too. At the end of the day, it's all going back to Google. This isn't to discount legitimate privacy and security issues with Google's AI features, as we saw recently with hackable Gmail AI summaries. However, it's equally important to note that not everything you see online is as dark and scary as it's sometimes made out to be.
