Thursday, September 24, 2026
Hoo-hoo, today Anthropic's AI put on a lab coat and found something scientists are comparing to Crispr, while Google teases its next big model and Meta ditches cameras on its new glasses.
Anthropic's AI Made a Lab Discovery It's Comparing to Crispr
Anthropic just opened a real wet lab, the kind with test tubes and microscopes, and put its AI model Claude to work in it. The company says Claude 'autonomously discovered' a new enzyme system similar to the machinery behind Crispr, the famous gene-editing tool that lets scientists cut and edit DNA. Importantly, Anthropic says humans are still supervising every step. Claude did not run the lab alone. This is the first big result from Anthropic's new biology lab, and it lands right as the company prepares to go public. If the discovery holds up under further testing, it would be an early sign that AI can genuinely speed up scientific discovery, not just summarize research papers about it.
What this means for you: AI is moving from writing your emails to potentially helping find real medical breakthroughs, but for now a human is still checking its work.
What this means for your business: If you're in biotech, health, or research, this is a signal that AI-assisted lab work is becoming real. Start thinking now about how you'd add human oversight and safety checks if you bring AI into scientific workflows.
Source: The Verge
Radar 01
Google's Next Big AI Model Is Almost Here
Google DeepMind's new chief, Koray Kavukcuoglu, said in his first public interview that Gemini 4 is in its final polishing stage. Google has been seen as trailing rivals like OpenAI and Anthropic on flagship model releases, so this next version carries a lot of pressure to prove Google can still compete at the top.
What this means for you: A major new AI model may reshape what Google search, your phone, and other Google products can do for you soon.
What this means for your business: If you're comparing AI vendors or locking in contracts, hold off on long commitments until you see how Gemini 4 stacks up against current options.
Source: The Verge
Radar 02
Meta Ditches the Camera on Its Newest Smart Glasses
At its Meta Connect event, Meta unveiled its first camera-free AI glasses, promising a lighter build and up to 12 hours of battery life. The move comes as smart glasses and other wearable devices with hidden cameras have drawn public backlash over surveillance worries. Meta also showed off Muse Charm, a new palm-sized standalone gadget for its Muse AI assistant.
What this means for you: Expect more AI wearables designed to ease privacy fears, even as tech companies keep pushing gadgets that listen and respond to you all day.
What this means for your business: If your company is exploring wearable AI tools for staff or customers, privacy-by-design features like no cameras could become a real selling point, not just a nice-to-have.
Source: The Verge
Radar 03
Microsoft Shuts Down AI-Powered Hacking Service That Hit 12,000 Accounts
Microsoft disrupted a platform called EvilTokens that used AI to make mass account break-ins faster and easier for criminals. Before it was shut down, the service had already compromised roughly 12,000 accounts, showing how AI tools are lowering the skill needed to run large-scale cyberattacks.
What this means for you: Criminals now have AI-assisted tools that make hacking accounts faster, so strong passwords and multi-factor authentication matter more than ever.
What this means for your business: Security teams should assume attackers are using AI to scale up attacks, and budget for detection tools that can keep pace with automated, AI-assisted threats.
Source: Ars Technica
Try This Today
Ask your team: if we gave an AI tool access to a real, high-stakes process at our company (like Anthropic did with its lab), what human checkpoints would we insist on before trusting the output?
Quick Hits
- ChatGPT's mobile app is getting voice-based agentic features. Pro and Plus users can now use a Work tab on their phone to hand off multi-step tasks by voice instead of typing. [1]
- Enveda raised $311 million to push more nature-derived, AI-designed drugs into clinical trials. The round values the biotech at $2 billion, with treatments in testing for skin conditions and weight-loss maintenance after GLP-1 drugs. [2]
- New research auditing AI agents that use outside tools found silent failures: cases where a tool call looks successful on the surface but has actually gone wrong, especially in biology-related AI workflows. A reminder that agentic AI needs closer watching, not just trust. [3]
Enjoyed today's read? Help me get this in front of more people like you.
Start every morning a little sharper.
Free. Takes 5 minutes to read. No spam.
