🚗 ChatGPT Comes to Your Car Dashboard

OpenAI is bringing ChatGPT to the driver’s seat with a new voice-first experience on Apple CarPlay. With just your voice, you can now interact with AI hands-free from your car’s dashboard using the ChatGPT app. A small update that could make everyday drives a lot more convenient. Let’s dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: OpenAI has launched a voice-first version of ChatGPT for Apple CarPlay, letting drivers use the AI hands-free on the road.

OpenAI has rolled out ChatGPT Voice for Apple CarPlay giving iPhone users access to the AI assistant directly from their car's dashboard screen. The feature works through the existing ChatGPT mobile app and requires iOS 26.4 or later on a supported iPhone. Once connected drivers can open the app via CarPlay and interact with it using only their voice. They can start fresh conversations or pick up existing ones without touching their phone. OpenAI announced the rollout on April 2 describing it as "the voice mode you know now available on the go." However the feature comes with notable limitations. It has no access to maps navigation tools location data or other apps. It cannot give directions or assist with route planning. OpenAI has also issued safety guidelines urging users to set up the app before driving and rely only on voice while in motion.

ChatGPT is the first AI to land inside Apple's new voice-based conversational app category for CarPlay, and that's not a small thing. For years, Siri was the only voice option in the car, and drivers largely had to accept its limits. Now Apple has cracked the door open, and OpenAI was quick to walk through it. Google's Gemini is already deeply embedded in Android Auto, Tesla has been running Grok since last year, and the broader in-car AI market is on track to nearly double by 2029. The car is becoming the next big battleground for AI assistants, and whoever feels most natural to talk to while driving could build a serious loyalty advantage. The fact that Apple still controls access through a strict approval process means this won't become a free-for-all, but it does raise a pointed question for Siri: how long before drivers stop thinking of Siri as the default at all?

Viktor is an AI coworker that lives in Slack, right where your team already works.

Message Viktor like a teammate: "pull last quarter's revenue by channel," or "build a dashboard for our board meeting."

Viktor connects to your tools, does the work, and delivers the actual report, spreadsheet, or dashboard. Not a summary. The real thing.

There’s no new software to adopt and no one to train.

Most teams start with one task. Within a week, Viktor is handling half of their ops.

Add Viktor to Slack for free.

Evolving AI: Anthropic's accidental leak of Claude Code's source code has already been cloned over 8,000 times on GitHub, and takedowns haven't stopped the spread.

The Wall Street Journal reports that Anthropic has had over 8,000 copies and adaptations of its Claude Code source code removed from GitHub following an accidental leak. The code reveals the techniques Anthropic uses to run Claude as a coding agent, including a "dreaming" function that helps the model consolidate tasks. Competitors now have direct insight into how Claude Code is built and can use it as a blueprint to replicate its core capabilities.

What made containment nearly impossible was the speed of AI-assisted replication. One programmer quickly rewrote the leaked code in multiple languages so it remained available even as GitHub removed copies. This is a new kind of leak problem, one where the damage compounds faster than any legal or technical response can keep up. The timing couldn't be worse for Anthropic. The company is targeting an IPO at a $380 billion valuation and this incident follows a separate human-error leak about its upcoming Claude Mythos model just days earlier. Two significant leaks in quick succession raise real questions about internal information security at one of the most closely watched AI companies in the world.

Claude Code has grown into a $2.5 billion run-rate product with enterprise adoption making up 80% of revenue, and the company is gearing up for a $380 billion IPO valuation. At that scale, a packaging error that hands rivals a detailed feature roadmap and a window into how you build a production-grade AI coding agent is a real problem. A Bloomberg report has a senior Anthropic executive blaming "process errors" tied to the company's fast release cycle, which is an honest answer, but not a reassuring one. Anthropic sells itself as the safety-first AI lab, and two significant internal leaks in under a week start to chip away at that identity. The deeper issue is what this signals about how AI companies manage IP in a world where a leaked zip file gets mirrored thousands of times before anyone notices it's gone.

Evolving AI: Unlock Your Next Income Stream: 100 Side Hustles for Ambitious Founders.

In a world where financial freedom feels like a distant dream, smart people are building wealth on their own terms.

Don't let another month slip by watching others build their empire. Your next income stream is hiding in our database, waiting to be discovered.

👉 Download the Side Hustle Database Now

Evolving AI: Google has released Gemma 4, its most capable open model family yet, now fully available under the Apache 2.0 license for the first time.

Google's Gemma 4 is a family of four open models covering a wide range of hardware and use cases. The lineup includes an Effective 2B and 4B for edge devices and a 26B Mixture-of-Experts and 31B Dense model for more powerful setups. All four support vision inputs and agentic workflows out of the box. The 31B model currently sits 3rd on the Arena AI Text Leaderboard among open models and scores 85.7% on the GPQA Diamond scientific reasoning benchmark, second best among all open models under 40 billion parameters according to Artificial Analysis. The 26B MoE model activates only 3.8 billion parameters during inference making it fast and efficient at generation. Both large models fit on a single H100 GPU and can run quantized on consumer hardware.

For the past year, the open-source AI leaderboard has quietly become a Chinese affair. DeepSeek's surprise showing in early 2025 kicked things off, and by the end of the year Chinese models like Qwen and GLM had gone from a tiny sliver of global usage to nearly 30% of all self-hosted deployments, with Qwen even edging out Meta's Llama as the most-used open model worldwide. Meta's reputation took a hit when developers started questioning whether Llama's custom license was truly open at all. Into that gap steps Google with Gemma 4, and the Apache 2.0 license is arguably the bigger story here. It removes the legal grey area that haunted earlier Gemma versions and gives developers something they actually want: certainty. Build with it, ship with it, modify it with no surprises down the road. Whether it can hold its ground against the relentless pace of Chinese open-weight releases is the real test ahead.

This video looks back at the moment AlphaGo shocked the world and changed how we think about AI. What started as a game breakthrough quickly turned into real-world impact, from science to drug discovery. A watch that shows why that one match still shapes today’s AI race.

QUICK HITS

🎙️ OpenAI acquires TBPN.

💰 Microsoft to invest $10 billion in Japan for AI and cyber defence expansion.

📈 Microsoft announced 3 new world class MAI models.

⚠️ New York AG warns of AI tax scams as tax day approaches.

📈 Alibaba's Qwen launches new flagship LLM with Qwen 3.6-Plus.

 *partner link