Hey there,

We've got some exciting AI news today.

Claude can now click buttons and fill forms in Chrome. Anthropic just settled its first major copyright case. And Google's new AI weather model is predicting hurricanes better than traditional systems.

Let's dive right in.

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

  • Get daily AI news, tools, and tutorials

  • Learn new AI skills you can use at work in 3 mins a day

  • Become 10X more productive

1. Claude for Chrome Lets AI Control Your Browser

Anthropic is piloting Claude for Chrome with 1,000 Max plan users, allowing the AI to take actions directly in your browser like clicking buttons, filling forms, and navigating websites on your behalf. Without safety mitigations, the attack success rate from malicious actors was 23.6%, but with new defenses including site permissions and action confirmations, they've reduced this to 11.2%. The extension requires users to grant specific website access and confirms before high-risk actions like purchasing or sharing data, using improved system prompts and blocking high-risk categories like financial services.

2. Google Launches Gemini 2.5 Flash Image for State-of-the-Art Image Generation

Google introduced Gemini 2.5 Flash Image (aka nano-banana), their new image generation model that can blend multiple images, maintain character consistency across prompts, and make targeted transformations using natural language. The model is priced at $30 per million output tokens ($0.039 per image) and is available now via the Gemini API, Google AI Studio, and Vertex AI. It can generate 50 possible scenarios for complex edits, place the same character in different environments, and benefits from Gemini's world knowledge to understand hand-drawn diagrams and follow complex editing instructions.

3. Google Translate Adds Live Conversation Mode with AI

Google launched real-time live translation in the Translate app for over 70 languages, allowing back-and-forth conversations with audio and on-screen translations that intelligently identify pauses and accents. The app also now offers customized language learning with AI-generated practice sessions that adapt to your skill level, available for English speakers learning Spanish and French. Using advanced voice recognition models trained to isolate sounds, the system works even in noisy environments and is initially available for users in the US, India, and Mexico.

4. Anthropic Settles Class Action Copyright Lawsuit from Authors

Anthropic reached a settlement with authors who sued over allegedly downloading 7 million pirated books to train Claude, marking the first major settlement in AI copyright cases. A California judge had ruled that while training AI on copyrighted works might be fair use, downloading from pirate sites could make Anthropic liable for billions in damages at up to $150,000 per work. The settlement terms weren't disclosed, but author's attorney Justin Nelson called it "historic" and said it will benefit all class members, with details coming in weeks.

5. Google's AI Model Predicts Hurricane Erin's Path Better Than Traditional Forecasts

Google DeepMind's new experimental cyclone model can predict hurricane formation, track, intensity, and size up to 15 days in advance, generating 50 possible storm scenarios compared to traditional models' 3-5 day forecasts. In testing, the AI model's 5-day track predictions were on average 140km closer to actual locations than leading physics-based models, equivalent to a 1.5-day improvement that typically takes a decade to achieve. The National Hurricane Center is now using these AI predictions alongside traditional models for the 2025 Atlantic hurricane season through Google's new Weather Lab platform.

6. Cloudflare Launches MCP Server Portals to Secure AI Connections

Cloudflare released MCP Server Portals in open beta, providing a centralized gateway to secure Model Context Protocol connections between LLMs and applications like Slack, Jira, or internal databases. The system addresses critical security vulnerabilities after incidents like CVE-2025-6514 exposed MCP authentication and the "NeighborJack" attack found hundreds of servers exposed to the internet. By routing all MCP traffic through Cloudflare, organizations get centralized policy enforcement, comprehensive logging, and curated access based on least privilege principles, preventing unauthorized AI agents from accessing sensitive systems.

Take your AI skills deeper 🚀

In AI Mastery Lab, we turn AI & Automation into real systems. Our free community shares daily automations, proven prompts, and helps each other build things that actually work.

Plus instant access to all my prompts, automations, and resources.

How would you rate today's newsletter?

Vote below to help us improve the newsletter for you.

Login or Subscribe to participate

Stay tuned for more updates, and have a fantastic day!

Cheers,
Zephyr

Keep Reading

No posts found