In partnership with

The Lobster AI Assistant That Went Viral — and Might Run Wild on Your Computer

Key Takeaway: Moltbot (formerly “Clawdbot”) is a viral, open-source personal AI assistant built by Peter Steinberger that can actually perform actions across your digital life—but that power comes with serious security risks, especially prompt-injection attacks.

More Insights:

• Started as one developer’s personal project; now exploded with early adopters who love tinkering.

• Renamed from Clawdbot to Moltbot after a legal challenge tied to Anthropic/“Claude” branding.

• Promises real-world automation: calendar management, app messaging, flight check-ins, and more.

• Skyrocketed to 44,200+ GitHub stars, with hype so strong it allegedly stirred investor buzz around Cloudflare.

• Core danger: “doing things” can mean executing arbitrary commands—making malicious messages a potential attack vector unless sandboxed.

Why it matters: Tools like Moltbot show what “agentic AI” looks like when it stops chatting and starts acting—and that shift forces a new reality: the biggest limiter won’t be model intelligence, it’ll be whether we can build trust, containment, and safeguards strong enough to let autonomy exist without turning everyday life into an exploit surface.

Better prompts. Better AI output.

AI gets smarter when your input is complete. Wispr Flow helps you think out loud and capture full context by voice, then turns that speech into a clean, structured prompt you can paste into ChatGPT, Claude, or any assistant. No more chopping up thoughts into typed paragraphs. Preserve constraints, examples, edge cases, and tone by speaking them once. The result is faster iteration, more precise outputs, and less time re-prompting. Try Wispr Flow for AI or see a 30-second demo.

Start flowing free

Anthropic, Apple, OpenAI CEOs condemn ICE violence, praise Trump

Key Takeaway: Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and Apple’s Tim Cook condemned alleged ICE/Border Patrol violence tied to Minneapolis—yet each paired their concern with praise for President Trump, fueling backlash from employees who want stronger, public action.

More Insights:

• Amodei publicly criticized the “horror” in Minnesota, emphasized defending democracy at home, and said Anthropic has no ICE contracts.

• A leaked internal Slack message shows Altman saying “ICE is going too far,” while also calling Trump a “strong leader” and hoping he “rises to the moment.”

• A leaked Cook email called him “heartbroken,” noting a “good conversation” with Trump—prompting reported anger from Apple employees.

• Tech workers and organizers are pressuring CEOs to demand ICE leave U.S. cities and to cancel any ICE-related contracts across Big Tech.

• The piece argues these careful statements reflect high stakes: the administration’s AI-friendly posture has coincided with massive fundraising and growth for OpenAI and Anthropic.

Why it matters: This is a live stress test for corporate power in a political moment—when companies benefit from government policy, even “speaking out” can become a calibrated PR exercise, revealing whether leadership’s values are convictions or just risk management.

Chrome Just Got Its Own “AI Agent” — and It Wants to Shop the Web for You

Key Takeaway: Google is upgrading Chrome with deeper Gemini sidebar integration and new agentic tools that can understand tab groups, edit images, and potentially complete online tasks autonomously using your personal context.

More Insights:

• Gemini moves from a floating window to a persistent sidebar that can answer questions about the current site and other tabs.

• Chrome can treat related pages as a “context group” when you open multiple tabs from one source—useful for comparisons like shopping.

• The sidebar is expanding beyond Windows/macOS to Chromebook Plus users.

• A personal intelligence layer is coming that can pull from Gmail, Search, YouTube, and Photos—enabling things like asking about schedules or drafting/sending emails from Chrome.

• The biggest bet is auto-browse, an agent that can navigate websites to do tasks (like buying items and hunting coupons), pausing for confirmation on sensitive steps like logins and purchases.

Why it matters: Google isn’t just adding an assistant—it’s turning the browser into an operating layer for your digital life, where “clicking around” becomes optional. If it works reliably, Chrome could shift from being a tool you use to a tool that acts for you—but the real battle will be trust, privacy, and whether agents can survive the messy reality of the web.

Other stuff

• Inside OpenAI’s big play for science

• SoftBank in Talks to Invest Up to $30 Billion More in OpenAI

• AI Discovered 12 out of 12 OpenSSL Vulnerabilities

• Trump’s acting cybersecurity chief uploaded sensitive government docs to ChatGPT

• ChatGPT Containers can now run bash, pip/npm install packages, and download files

• Google Photos’ latest feature lets you meme yourself

• OpenAI Wants To Create Biometric Social Network To Kill X’s Bot Problem

• Will Your AI Teammate Bring Bagels to Standup?

Hope you enjoyed today's newsletter

⚡️ Join over 300,000 people using the Superpower ChatGPT extension on Chrome and Firefox.