Thursday, May 9, 2024

Welcome to TechCrunch AM! After years of not getting COVID, the COVID has come for me. But that doesn't stop the news! So, this morning, we have notes from OpenAI about how its models work behind the scenes, a possible IPO in 2025 from a Thai unicorn, several cybersecurity stories, and even a quick look at Robinhood's earnings. Let's go! — Alex

TechCrunch Top 3

Image Credits: Bryce Durbin / TechCrunch

1. OpenAI's secret(ish) AI instructions: Tired of ChatGPT telling you that it can't do something? Want to know the real rules in the background that you are interacting with but cannot see in modern AI models? OpenAI is pulling the curtain back somewhat so that we can better understand what's going on beneath the hood. Read More

2. Everyone hates Apple's iPad ad: What if you crush tons of artistic supplies and offer to replace instruments, brushes and paint and the like with a single, flat computer? Apple probably thought that would be a slick way to show off its new tablets. Lots of folks disagree. Read More

3. Thai food delivery app considers US IPO: If you are shareholder in Line Man Wongnai, good news! The company may list in the United States next year. The news matters for two reasons: First, the company's 2022 Series B made it a unicorn. Second, the United States remains the place to list if you are so inclined. Read More

Morning Must Reads

Image Credits: NurPhoto / Contributor / Getty Images

Encryption is not enough to protect you: The Spanish police have apparently identified a Catalan activist who used encrypted services Wire and Proton, as well as Apple. While the contents of the activists' messages were safe due to encryption, the police managed to secure metadata, and thus find out who they were. Encryption is amazing, but metadata is what it all comes down to, it appears. Read More

Hey look, the government can't stop leaking: The U.S. Patent and Trademark Office said this week that trademark applicants had their private domicile addresses appear in public records from August 2023 through April of this year. This breach underscores my belief that cybersecurity is the modern corollary to Sysiphus: We are going to keep fighting, but never complete the task of securing our digital lives. Read More

India could delay imposing caps on UPI providers' market share: That's good news for Google and PhonePe, which offer payment services through India's Unified Payments Interface (UPI). There's concern that the two companies control too much of the market, but the country's payments regulator is considering not imposing caps for at least a year or two, TechCrunch's Manish Singh reports. Read More

Around the Web

Everything is copyright: Stack Overflow is banning users who want to delete or avoid making their content on the platform for use by OpenAI. Stack Overflow is working with OpenAI "to scrub the site's forum posts to train ChatGPT," Tom's Hardware reports. Seems like everything is going just fine over there. Read More

Robinhood crushes earnings: Equity trading service Robinhood had a great Q1, handily beating analysts' expectations for revenue and swinging to a profit, CNBC reports. The company's shares are up 5% in trading before the bell this morning. Read More

Mistral cooks up a tailwind: French AI model company Mistral is looking to raise $600 million at a valuation of $6 billion, the Wall Street Journal reports. That's a lot of capital at a massive price tag — the company may be Europe's best hope for an AI champion. Read More

A MESSAGE FROM INCOGNI

Stop Robocalls, Spam Emails, Stalkers, Frauds, and More

Every day, data brokers profit from your sensitive info—phone number, DOB, SSN—selling it to the highest bidder. And who's buying it? Best case: companies target you with ads. Worst case: scammers and identity thieves. Try Incogni—not only does it delete your personal data from the web, but it also removes your info from sites used for stalking.

Before You Go

Image Credits: Jakub Porzycki/NurPhoto / Getty Images

TikTok to mark AI-generated content as artificial: AI-generated content made on other platforms is going to get narc'd on by TikTok — the company is starting to automatically label such content as made by AI. The move could limit the impact of deepfakes and other confusing content that might mislead the masses. Read More