Scale AI just laid off 14% of its workforce: read the full email the interim CEO just sent employees
This comes after Meta invested $14 billion into Scale AI last month as part of a blockbuster AI deal that included the hiring of Scale's ex-CEO, Alexandr Wang, to lead its new Superintelligence group and the purchase of equity in almost half the data labeling startup.
The company is 'restructuring' its GenAI group according to an email from Scale's interim CEO, Jason Droege, obtained by Business Insider. The GenAI organization is a core part of Scale that manages its work on some of the best-known AI chatbots like xAI's Grok and Google's Gemini.
The move was abrupt, one source told BI, with employees shut out of systems and deactivated from Slack even before they woke up.
Questions about Scale AI's long-term sustainability swirled, with some of Scale's biggest clients like Google abruptly halting projects in the wake of the deal.
Scale remains well-funded and is committed to growth and hiring hundreds more employees in different sectors, the startup told BI.
"As a result of this restructuring, some members of this org will be leaving Scale today," the email reads. "To ensure a smooth process and give people the necessary space, please do not come into the office."
The source told BI that the Meta investment weakened Scale AI's core leadership, with not only Alexandr Wang departing but also key VPs, chiefs of staff, and researchers.
Laid-off ScaleAI employees will be paid as normal through September 15, and then receive a minimum of four weeks of additional pay if they sign the company's severance agreement, per a separate email to the affected employees obtained by BI.
Here is the full email from Scale AI's CEO Jason Droege:
Hi team,
After reviewing our company and its operations, I've made the decision to restructure several parts of the GenAI organization. I believe these changes will best position us for the long-term, make the org more efficient, and allow GenAI to focus on the biggest and most profitable opportunities.
As a result of this restructuring, some members of this org will be leaving Scale today. To ensure a smooth process and give people the necessary space, please do not come into the office. All impacted FTEs will receive an email to their personal email address by 8am PT and badge and IT access will be removed. For those not impacted, I will share more details later this morning on the restructuring and our path forward.
I know news like this is tough, but please know how grateful I am for the work this team has done. Whether you're directly affected by these changes or not, every one of you has helped build Scale into the company it is today and the accomplishments it has made.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
44 minutes ago
- Yahoo
Google Now Lets You Create Illustrated AI Storybooks Inside The Gemini App
Google isn't sitting idly by while OpenAI is busy releasing its newest open-source ChatGPT models and teasing the big GPT-5 upgrade. For example, Google added several AI agents to Google Cloud that can help app developers and businesses create faster and smarter apps. The company also released a mind-blowing new AI world model called Genie 3 that lets you interact AI-generated content. Finally, Google unveiled a new Storybook mode for Gemini that lets you create illustrated stories for your kids on any topic. The current AI models can't achieve anywhere close to the same level of creativity as humans. They can answer questions and solve problems, but creativity isn't their strong suit. When it is, it will likely be the dawn of artificial general intelligence (AGI), the next generation of AI that will give the AI the creativity it needs to address any situation like a human would. But the new Storybook mode in the Gemini app doesn't need the AI to be especially creative. It might be just what some parents need to keep their children entertained with a new bedtime story when they get bored of their favorites. It can also be a great tool to come up with new stories on the fly when your kid demands it, or to explain more complex concepts in a more digestible format. And Storybook doesn't just create a unique digital book by following a set of instructions -- it also provides picture book illustrations. Read more: ChatGPT Has A Built-In 'Hack' That Makes Your Prompts So Much Better How Storybook Works In The Gemini App Google explained in a blog post that the new Storybook mode is available globally on desktop and mobile in all languages that Gemini is available. The Storybook mode is available as a Gem or a customized chatbot. Gems are the equivalent of custom GPTs in ChatGPT. Google still labels the Storybook Gemini Gem as an experiment, warning that it may have unexpected results. This also signals that Google should improve it in the future. As for those unexpected results, you can expect hallucinations (which aren't bad for a story) and illustration issues. For example, The Verge was presented with an image of a fish with hands. But how else would the fish push a boulder? Other than that, the Storybook Gem lets you use Gemini as you normally would. You get to pick the AI model you want to use for the story, and you can upload files to customize the illustrated 10-page book that Gemini will imagine for you. The video above shows a Google demo of an AI story made with Storybook based on a prompt asking the AI to create a coloring book inspired by the mother's career. Choosing the type of illustration might be the best part of Storybook mode. For example, you can instruct the AI to use the style of your kid's drawings to illustrate the story. The examples that the Storybook page provides also show the AI will support Anime and Claymation art. You can tweak the book until you get the desired results. Read the original article on BGR.
Yahoo
an hour ago
- Yahoo
Claude AI Can Now Search And Reference Past Conversations
Claude is one of the main AI competitors to ChatGPT and Gemini, but Anthropic's chatbot doesn't necessarily offer feature parity with its rivals. For example, Anthropic released web search, deep research support, and voice mode in previous months, features that had been available in ChatGPT well before they made it to Claude. Fast-forward to mid-August, and Claude received a big memory upgrade that allows the chatbot to reference past chats. The feature is similar to a ChatGPT capability that allows the AI to reference past chats. Claude's "Search and reference chats" feature started rolling out on Monday across web, desktop, and mobile. Anthropic detailed the new feature in an updated support document that explains how Claude personalization works. The AI's ability to search past chats and reference those conversations isn't identical to ChatGPT's broad memory feature that can remember things about the user. Claude's chat memory isn't persistent, as it only works on a chat-by-chat basis. Read more: 5 Expert Tips To Get The Most Out Of ChatGPT Every Time You Use It How To Enable Claude's Search And Reference Chats Feature Anthropic on Monday started rolling out the new "Search and reference chats" feature to Claude Max, Team, and Enterprise users. The AI firm said it will expand it to other plans "soon." The example in the YouTube video above shows how Claude's chat memory feature will work. A user asks the AI what they were working on before their vacation, and Claude performs a search to identify the conversations that qualify. Claude discovers five results, listing them in the chat and providing a summary to the user so they can decide what topic to resume. To enable Claude's "Search and reference chats" feature, you'll have to go to the "Settings" menu in the app, click on "Profile," and find the "Preferences" section. In there, you'll have to enable the "Search and reference chats" feature by clicking the toggle. You can toggle off the feature if you want to prevent Claude from searching and referencing past chats. How Claude's Chat Memory Works Anthropic explains in the support document that Claude can perform searches in your chat history to find relevant conversations with the AI that can add context to the current chat and "keep your conversation flowing." However, Claude will not remember things about you like ChatGPT's broader memory feature does. Anthropic spokesperson Ryan Donegan informed BGR that Claude's new search functionality isn't persistent memory. Claude will not build a user profile. It will only perform searches when you instruct it to. Claude can search for any chat outside of projects or search within projects. In the second scenario, searches are limited to that project. Anthropic also explains that users can't exclude distinct chats from the search feature. The only way to ensure that specific conversations with the AI do not resurface in Claude searches is to delete them. You can disable "Search and reference chats" to prevent Claude from referencing chats you don't want the AI to find. But this method will not let you look for any old chats you might need. Comparatively, ChatGPT won't let you enable its "Reference chat history" feature that matches Claude's new memory ability without also turning on the "Reference saved memories" feature. The latter is OpenAI's broader memory feature for its chatbot. Also, OpenAI made available the memory features to ChatGPT Free users a few months ago. Read the original article on BGR.
Yahoo
an hour ago
- Yahoo
Google Gemini Chats Just Got More Personal
GPT-5 might be the most important thing happening in AI tech this month, but Google is also constantly upgrading the abilities of its Gemini chatbot. The company announced new Gemini features on Wednesday that users will surely appreciate. Gemini chats will get more personal than before, as the chatbot will be able to provide more personal answers thanks to a new "Personal Context" feature. The feature will roll out gradually, starting with Gemini 2.5 Pro users. Gemini is also getting more private, with Google rolling out support for Temporary Chats, a feature that's similar to how private browsing or private Google Maps navigation works. On top of that, Google will also make changes to the way you handle Gemini privacy. The "Gemini Apps Activity" section will be renamed to "Keep Activity" in the near future. It'll include settings that manage Gemini Live data from your iPhone or Android handset, including audio and recordings, just like the "Gemini Apps Activity." Read more: How To Remove Yourself From Google Results And Any Other Websites Gemini Personalization Is Turned On By Default AI firms like OpenAI and Google are improving the abilities of their chatbots with every new update. The goal isn't to just offer a more reliable assistant that can answer quickly all sorts of prompts. AI firms also want to create personal assistants that know everything about the user and can provide more helpful context. Personalization features, like remembering previous chats and specific memories, are stepping stones towards that future. Google on Wednesday said that Gemini will be able to reference past chats to learn what you like and offer personal responses. Here's an example of a personal Gemini experience from Google's announcement: "You've previously discussed the evolution of characters' powers in your favorite comic book. Now, if you ask Gemini to brainstorm a birthday party theme that's unique to me, it might suggest a celebration based on your favorite character, complete with themed food and a custom photo booth with props." The feature is called "Personal Context," and it's turned on by default in the settings menu of the Gemini app. The screenshots above show your customization options on mobile and desktop. You can disable "Your past chats with Gemini" if you don't want the AI to retain memories. Personalized conversations will be available with the Gemini 2.5 Pro model initially in select markets. It will roll out to Gemini 2.5 Flash in the coming weeks. Temporary Chats And New Privacy Controls Temporary Chats let you exclude Gemini conversations from your history. They won't appear in the list of recent chats, and won't impact Gemini's "Personal Context" memory feature. Also, Temporary Chats will not send data to Google to improve the Gemini if you have that setting enabled. To use Temporary Chats, you'll want to click on the chat bubble that appears on the right of the "New chat" menu. Temporary Chats will be retained for 72 hours, which is the standard for all Gemini chats. This is a safety and feedback feature. If you have the "Gemini Apps Activity" setting enabled in your Gemini apps, Google will use information from your chat to train the model. That's how other AI models work, including ChatGPT. But users can always opt-out, to exclude personal chats from the training data Google uses to improve its AI models. The "Gemini Apps Activity" feature will be called "Keep Activity" in the coming weeks. Despite the name change, the purpose of the setting will not change. Turn it off, and Google won't train the AI with your data. If "Gemini Apps Activity" is off, "Keep Activity" will remain off. Google also explained that it introduced a privacy feature earlier this month that lets Gemini Live users decide whether audio, video, and screenshots can be used to train Gemini. That feature is off by default, so you don't have to do anything to prevent your data from reaching Google's servers. As you can see above, the upcoming "Keep Activity" setting will also include a tick box for audio and Gemini Live recordings. Do not tick the box if you want to prevent Google from using that data to improve its AI models. Read the original article on BGR.
