Scale AI used public Google Docs for confidential work with Meta, xAI in stunning revelation after $14B investment: report
The artificial intelligence start-up that recently clinched a $14 billion investment from Meta has an 'incredibly janky' security system – using public Google Doc files to store confidential information on clients like Meta, Google and xAI, according to a report.
It was reported earlier this month that Meta agreed to take a 49% stake in Scale AI for $14.8 billion and bring the startup's CEO Alexandr Wang over to lead a new 'superintelligence' lab.
That shockingly steep price tag indicates that Meta believes Wang and his company are key to bringing the social media firm's AI division to the next level.
Advertisement
3 Scale AI CEO Alexandr Wang will be leading Meta's 'superintelligence' lab as part of the deal.
REUTERS
But the company has been strangely relaxed when it comes to its work with high-profile clients, leaving top-secret projects and sensitive information like email addresses and pay details in Google Docs accessible to anyone with a link, according to Business Insider.
'We are conducting a thorough investigation and have disabled any user's ability to publicly share documents from Scale-managed systems,' a Scale AI spokesperson told BI.
Advertisement
'We remain committed to robust technical and policy safeguards to protect confidential information and are always working to strengthen our practices.'
While there is no indication the public files have led to a breach, they could leave the company susceptible to hacks, according to cybersecurity experts.
Scale AI, Google and xAI did not immediately respond to The Post's requests for comment. Meta declined to comment.
Five current and former Scale AI contractors told BI that the use of Google Docs was widespread across the company.
Advertisement
3 Several 'confidential' documents for client Google were publicly available as Google Doc links, according to the report.
REUTERS
'The whole Google Docs system always seemed incredibly janky,' one worker said.
BI said it was able to view thousands of pages of project documents across 85 Google Docs detailing Scale AI's sensitive work with Big Tech clients, like how Google used OpenAI's ChatGPT to fine-tune its own chatbot.
At least seven Google manuals marked 'confidential' including recommendations to improve the chatbot, then-called Bard, were left accessible to the public, according to the report.
Advertisement
Public Google Doc files included details on Elon Musk's 'Project Xylophone,' like training documents with 700 conversation prompts to improve an AI chatbot's conversation skills, the report said.
So-called 'confidential' Meta training documents with audio clips of 'good' and 'bad' speech prompts to train its AI products were also left public.
3 Some 'confidential' documents related to Meta's training of AI products were also publicly available, the report said.
REUTERS
While these secret projects were often given codenames, several Scale AI contractors said it was still easy to figure out which client they were working for.
Some documents tied to codenamed projects even mistakenly included the company's logo, like a presentation that included a Google logo, according to BI.
When working with AI products, sometimes the chatbot would simply reveal the client when asked, the contractors said.
There were also publicly available Google Doc spreadsheets that listed the names and private email addresses of thousands of workers, the news outlet found.
Advertisement
One spreadsheet was frankly titled 'Good and Bad Folks' and labelled dozens of workers as either 'high quality' or 'cheating,' BI said. Another document flagged workers with 'suspicious behavior.'
Still more public documents laid out how much individual contractors were paid, and included detailed notes on pay disputes and discrepancies, according to the report.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
31 minutes ago
- Yahoo
What to do if your tablet is lost or stolen: tips to help ease your concerns
Tablets are the go-to for entertainment on travels for adults and kids alike, but they contain much more than the latest episode of The Traitors, with access to the ins and outs of daily life. So when one gets lost or stolen, the cost of the slate isn't the only thing to worry about. If yours disappears, here's what to do. Try to locate your tablet with Find My on Apple or Google, or Manage Content and Devices for Amazon's Fire tablets. Remotely lock your tablet using Find My or management settings. Here you can mark it as lost, block the use of Apple or Google Pay, make it play a sound and leave a message on the screen for anyone who finds it. You can also remotely erase your tablet if you can't find it. If you have mobile data on your tablet, contact your mobile broadband provider and block your sim to stop thieves running up bills. Contact your credit card company for any cards you have stored on your phone and disable Apple or Google Pay. Report the theft to the police on 101 and give them your phone's IMEI number, which may be on the box, in your Amazon, Apple or Google account or Find My services. Contact your insurance company if you have tablet cover. Report the theft to your tablet's manufacturer so they can flag it as stolen next time it connects to the internet. Change your passwords for key accounts. Start with your email account so thieves can't gain access to your other accounts through password resets. Deregister your tablet and remove it from your various accounts and services, which will log it out and stop thieves accessing saved details. When you get a new or replacement tablet there are some steps you can take to help if it goes missing. Set a strong pin, set a short screen-lock timeout and turn on biometric fingerprint or face scanners if your tablet has them. Turn on Find My or location tracking on your tablet in the settings, which allows you to locate it, lock it or erase it remotely via a web browser or another device. Take a note of your tablet's serial number, which can be found on the box or in system settings. Use biometrics for any banking and sensitive apps that support them, to block access. Disable access to quick settings, Alexa, Siri or Google Assistant/Gemini and notifications when your tablet is locked, which prevents thieves turning off internet access or accessing some of your data. Back up your tablet using iCloud on an iPad, Google Drive on an Android tablet or Amazon's Backup and Restore feature on a Fire tablet.
Tom's Guide
38 minutes ago
- Tom's Guide
I actually prefer budget phones to flagships — but there's one thing phone makers need to fix
I'm a big fan of phones will small price tags. That's partly a reflection of my inherent cheapness — you'll have to use a crowbar to pry a dollar bill free from my wallet — but it's also an acknowledgment of a simple truth: the very best cheap phones can do nearly everything a flagship device can do these days. That's not to say you won't have to make compromises in some areas. Screamingly fast processors are generally reserved for more expensive phones, though unless you're regularly using demanding apps or graphically intensive games, you'll likely be just fine with the good-enough chipsets found in cheaper phones. More expensive models also benefit from better camera hardware and more extensive AI support, but there are enough lower-cost phones that hold their own in camera comparisons to make that trade-off less noticeable. As for AI, well, if you consider that feature a must-have for your phone, feel free to spend big on the latest flagships. For most people, though, AI-powered capabilities are a nice-to-have, not an essential part of the smartphone experience. That said, there remains one area where cheaper phones are clearly at a disadvantage. And it simply doesn't need to be that way. Cheap phones need more extensive software support from their device makers. That's especially true in this era where people want to hold on to their phones for longer, for reasons relating to both budgets and the environment. Our pick for the best cheap phone available right now — Google's Pixel 9a — stands out as a model of what phones under $500 could offer their users. Buy a Pixel 9a, and Google promises to deliver seven years of software and security updates. Notably, that's the same level of support you'd enjoy if you paid up for a flagship Pixel. But the Pixel 9a is clearly an outlier when it comes to software support. Scour the other sub-$500 phones we recommend, and you're not going to find any matching Google's pledge. Phone Price Software updates Security updates Google Pixel 9a $499 7 years 7 years Samsung Galaxy A36 $399 6 years 6 years iPhone 16e $599 5 years (unofficial) 5 years (unofficial) OnePlus 13R $599 4 years 6 years Nothing Phone 3a Pro $459 3 years 6 years Nothing Phone 3a $379 3 years 6 years Moto G Power 2025 $299 2 years 3 years Moto G 2025 $199 2 years 3 years The Nothing Phone 3a and Nothing Phone 3a Pro are both fine low-cost alternatives to the Pixel 9a, particularly when you consider that Nothing equips its phones with dedicated telephoto lenses. But both devices are only set to receive 3 years of Android updates, along with an additional 2 years of security support. Samsung's Galaxy A series of midrange devices come the closest to matching what Google offers Pixel 9a buyers. A Galaxy A36 phone receives 6 years of software and security updates, as well the Galaxy A56 when that phone ships in the U.S. later this year. (It's currently available in other parts of the world with that six-year guarantee, though.) Motorola brings up the rear, as its latest Moto G phones feature 2 years of Android updates and a third year of security support. That's actually a step up from the 2024 Moto G series, which only supported one Android update. Expand your search to phones that cost a little more than $500, and you'll see some more extensive support — though Google still sets the pace. The $599 iPhone 16e gets the same access to iOS updates as Apple's other iPhone 16 models, which should mean at least 5 years, though recent iPhone models have been able to count on longer support. The OnePlus 13R — a terrific midrange model — limits Android updates to 4 years, though you do get six years of security support with that phone. If you're buying a cheaper phone, it's likely because you don't want to spend $799 or more on a smartphone, which you'd have to do by turning to a flagship. But when phone makers cap the software support on a device, you're not getting as much value from your low-cost purchase as you should. Let's say you buy a Moto G 2025 — and why not since you get epic battery life, a stylish design and surprisingly good camera performance in a $199 device. But you'll also run into Motorola's limited software support, which means Android 17 is the last update you'll be able to count on. Maybe having the latest OS isn't as much of a priority to you, but even if you hold on to the Moto G until security updates run out, you'll still be looking for a new phone by 2028. And those seemingly low-cost purchases start to add up. There's more than just frugality at stake here. Even with phone makers using more recycled materials for their devices, it's generally a more earth-friendly approach to space out your phone upgrades even longer than you are now. Extending software support is a seemingly easy way to build more sustainable devices. I think phone makers are starting to recognize this. As noted Motorola's 2025 Moto G phones have better support than their predecessors, and that's not the only company moving in the right direction. Samsung used to cut off support for its Galaxy A phones at 4 years instead of 6. The OnePlus 13R's software support policy is more generous than what OnePlus offered for the OnePlus 12R. And both the Nothing Phone 3a and 3a Pro get an extra 2 years of security support compared to the Nothing Phone 2a. And Nothing has already confirmed that the Nothing Phone 3 is getting 5 years of Android updates plus seven years of security support. These are welcome changes, and I hope to see the trend continue with future phone releases. There should certainly be clear distinctions between flagship phones and more affordable models. But how long you're able to keep your phone shouldn't be one of those differences.
Tom's Guide
an hour ago
- Tom's Guide
I swapped the Calm app for ChatGPT Voice — here's what surprised me
Like a lot of people, I rely on calming apps to help me wind down after chaotic days. With three kids and a busy schedule, every day is chaotic for me. For years, the Calm app has been my go-to for sleep stories, meditations and mid-day stress breaks — but with AI voice assistants getting smarter, I started wondering: could ChatGPT Voice replace my Calm subscription and save me a little each month? I already use ChatGPT Voice as part of my ChatGPT Plus subscription almost daily for brainstorming ideas, quick research and productivity hacks. So I figured why not test it as a pocket-sized calming coach? For one week, I swapped out the Calm app and used ChatGPT Voice for meditation, stress relief and sleep. Here's what worked, what didn't and what I would recommend. There's no official 'calm mode' in ChatGPT, but with a few simple voice prompts, you can easily get it to guide meditations, breathing exercises or calming conversations. I didn't even need to use custom instructions. Once you pick the ChatGPT Voice, prompt the chatbot. Here are the prompts I used: Get instant access to breaking news, the hottest reviews, great deals and helpful tips. At first, the biggest challenge was that ChatGPT Voice doesn't have built-in soundscapes or pre-recorded sessions like Calm (unfortunately, no Harry Styles voiceover here). The first few sessions felt a little robotic, but by day 3, ChatGPT Voice was adapting to my preferences: slower pace, softer tone and more natural breathing cues. You can even ask the chatbot to whisper. I had to be a little more proactive with my requests, and the chatbot often stopped too soon, which made me prompt 'keep going.'However, I quickly found that the flexibility was a big plus. I could ask for exactly the kind of calming I needed in the moment, without browsing menus. I didn't have to rely on a meditation app template, and I could change the meditation on the fly. The first few sessions felt a little robotic, but by day 3, ChatGPT Voice was adapting to my preferences: slower pace, softer tone and more natural breathing cues. You can even ask the chatbot to whisper. By midweek, I started using ChatGPT Voice for quick 'reset' breaks between meetings and writing — something I rarely did with Calm, since it takes longer to navigate. I'd say: 'Let's do a quick mindful break,' and in seconds, I'd get a calming, 2–3 minute reset. Sleep support was a mixed bag — the voice is soothing, but without background sounds, I didn't fully relax the way I would with a Calm sleep story. Pros: More personalized than Calm — you can shape the tone and style No extra subscription (already part of ChatGPT Plus) Faster for on-demand calming moments Learns your preferences over time Cons: No built-in music, soundscapes, or polished narrators Slight learning curve to figure out the best prompts Not as effective for deep sleep as Calm's full app If you already use ChatGPT Plus, Voice can absolutely double as a casual calming app — and it's more flexible than you'd think. For those who rely heavily on the immersive sleep stories and ambient soundscapes of Calm, it's probably not a full replacement (yet). But for quick daytime resets, guided breathing, and moments of calm on the fly? I'm keeping ChatGPT Voice in my wellness toolkit.
