Thursday, March 20, 2025

Welcome to TC PM! Today we take a look at what's up with Tesla now that Elon Musk is focused on DOGE; we learn that Anthropic's Claude can surf the web; and we admit that AI is actually decent at determining whether art is authentic.

TechCrunch Top 3

Image Credits: Sean Gallup/Getty Images

🤭 Oops! Elon Musk has been spending too much time away from Tesla — at least according to one Tesla analyst. Wedbush Securities' Dan Ives, one of Tesla's biggest boosters, wrote a letter saying that Musk was facing a "moment of truth" and pleading with him to reduce his role with DOGE.

🛻 Speaking of Tesla: Tesla is recalling 46,000 Cybertrucks after the National Highway Traffic Safety Administration found that an exterior steel panel on the side of the truck's windshield can peel off while in motion. That's not good.

↗️ Upgrade time: OpenAI is releasing new transcription and voice-generating AI models to its APIs. The company claims they are improved versions of what was already there and that they hallucinate less.

A Message from Anthropic

Free Anthropic event: How AI is shaping the future for startups

Ready to move your AI projects from evaluation to impact? Join this practical discussion for startups and tech-forward companies that are ready to build and scale AI solutions that deliver measurable business results.

Afternoon must-reads

Image Credits: Getty Images

🤝 More M&A: German reinsurance giant Munich Re scooped up Next Insurance, a startup focused on insurance for small- to medium-sized businesses. The purchase price was $2.6 billion, which is a slight bump over the $2.5 billion valuation Next Insurance raised at in 2023.

🧊 Let's connect it: Data movement company Airbyte released a host of new features today, including compatibility with Apache Iceberg and support for moving unstructured data out of sources like Google Drive and SharePoint.

🕳️ ServiceNow vulnerabilities: Hackers are looking to exploit a trio of ServiceNow vulnerabilities — again. These vulnerabilities were first discovered by security researchers last May. ServiceNow patched them up in July, but now hackers are starting to swarm again.

🇪🇺 AI to the EU: Meta's AI-powered chatbot, Meta AI, is finally rolling out in the EU, despite an ongoing regulatory battle. The EU's version of Meta AI won't come with all the same features U.S. users see, though. For example, the AI won't be able to create images in the EU.

🏄 Claude surfs the web: Anthropic's AI chatbot Claude can now search the web. For now, web search only works with Claude's latest model, Claude 3.7 Sonnet, and is a preview for paying users. This new capability brings Claude on par with its AI chatbot peers.

Around the web

🎨 AI in art: It's unclear what role AI will end up playing in the future of art. While AI is definitely not an artist yet — nor do many people want it to become one — it is good at one thing: helping experts with art authentication.

☹️ What the? Megan Garcia, the mother who is suing Google and Character AI over their involvement in her 14-year-old son's suicide, discovered that AI chatbots that are based on her son are being hosted on the platform.

Before you go

Image Credits: Minecraft

🕹️ The kids will be all right: A high school student named Adi Singh built a website called Minecraft Benchmark, or MC-Bench. It pits AI models against each other in Minecraft to see which one can build better creations based on prompts.