Welcome back to The Prompt.
The explosion of generative AI has breathed new life into humanoid robotics, with buzzy companies like Figure AI emerging. But a small startup spun out of University of Texas in Austin has been quietly building humanoid robots for the past eight years. Apptronik’s bipedal robot, Apollo is able to carry out factory tasks like picking up heavy objects and moving them across warehouses. The proof? Mercedes has been testing the machines in a plant in Hungary. Now, the company is raising $100 million in venture capital to build out hundreds of more of the latest version of its Apollo robots.
Now let’s get into the headlines.
POLITICS + ELECTION
X’s AI chatbot Grok regurgitated false claims that former President Donald Trump was “missing dentures” during his interview with Elon Musk on X on Monday. The inaccurate claim came after multiple users remarked on X that they thought that Trump was slurring his words during the interview. Grok, which produces content based on trending topics on X, published misinformation in an AI-generated summary titled “Trump’s Dentures In Discussion Amid Speech.” The post was quickly removed from the platform.
This error came days after Trump falsely suggested that an image of Vice President Kamala Harris’ rally crowd in Michigan was an AI-generated fake.
PEAK PERFORMANCE
A new coding engine called Cortex, launched by AI startup Codeium, can process up to 100 million lines of code at once, CEO Varun Mohan told Forbes. That’s important because learning from more data can make the AI model’s suggestions more relevant and improves its efficiency in applying tweaks or fixes.
Codeium is one of 25 startups to be featured on this year’s Forbes Next Billion Dollar Startups list, a compilation of venture-backed companies that are most likely to reach a unicorn status. This year’s list is packed with AI companies like video generation startup HeyGen and AI development platform Fireworks AI.
AI DEAL OF THE WEEK
Legal tech startup EvenUp has raised $35 million in funding in a round led by Lightspeed Venture Partners. The startup’s AI tool, trained on 250,000 verdicts and settlements, helps personal injury lawyers spin up demand letters and other legal documents to obtain compensation from insurance companies for claims resulting from incidents like car accidents.
DEEP DIVE
Les Wexner, the Victoria’s Secret billionaire who stepped down from his company L Brands in 2020 amid scrutiny over his ties to convicted sex offender Jeffrey Epstein, is cashing in on the AI surge.
Thanks to a $1 million seed investment in Atlantic Crypto, a little-known company that would become AI cloud juggernaut CoreWeave, the 86-year-old’s family trust now holds a $720 million stake in one of the most valuable AI startups around. The Ohio-based billionaire is one of America’s wealthiest people, worth an estimated $6.3 billion.
CoreWeave, which provides access to highly coveted chips used in building AI models, is among the most valued AI startups to emerge from the mainstreaming of artificial intelligence. It raised $7.5 billion in debt from Blackstone in May, and $1.1 billion in equity earlier this year to build out a massive data center network that delivers the computing power needed to train AI. But back in 2019 it was a struggling cryptocurrency miner that had just started subletting its graphic chips to AI startups from a New Jersey garage.
CoreWeave raised $1.2 million in a seed round in March 2019, according to Pitchbook data. Wexner’s money manager Greg Hersch of Florence Capital invested $1 million then, later picking up another $600,000 in stock from the company’s Series A round and doubling down again through an additional convertible note. All of this was placed in a trust established for the benefit of Wexner’s four children: Sarah, Hannah, David and Harry.
Now, those investments have evolved into a $720 million stake in the AI startup that’s rumored to be seeking an initial public offering next year — and a legal squabble over what appears to be an accidental jackpot.
Read the full story in Forbes.
YOUR WEEKLY DEMO
Google said it’s making it easier for people to request removal of AI-generated explicit images of themselves from showing up in search results. The company is updating its policies so that when someone successfully requests removal of an image, Google will also filter results on similar searches and its systems will also look for and take down duplicates of those deepfake images. The tech giant also said sites that have received a high number of removal requests for fake explicit imagery will be demoted in search results.
AI INDEX
50%
Job applicants are using AI to generate résumés and cover letters, according to estimates from recruiters and surveys, The Financial Times reported.
QUIZ
What company has AI unicorn Hugging Face acquired to host advanced and large-scale AI models?
- MosaicML
- XetHub
- CharacterAI
- Argilla
Check if you got it right here.
MODEL BEHAVIOR
When Humane’s $699 AI pin launched in April, it was met with brutal reviews. Customers who had bought the wearable device —which was supposed to do everything a phone could, like take photos and answer questions — are now realizing that they don’t want it. Daily returns of the device are outpacing sales, The Verge reported. About $1 million worth of products have been returned to the startup, which is backed by heavyweights like OpenAI CEO Sam Altman and Salesforce CEO Marc Benioff.
