In partnership with
Welcome, busybees
An Anthropic staff member with no serious math training asked an unreleased Claude model to "take a real stab" at the Riemann Hypothesis, one of math's most famous unsolved problems with a $1 million bounty still unclaimed, then let it run largely on its own for about a day and a half. What came back was real: the model raised a longstanding lower bound tied to the hypothesis from 41.6% to 67.2%, coordinating 60 subagents through 650 different approaches and 31 million output tokens, with two Anthropic mathematicians independently verifying the result and a formal Lean proof produced alongside it. Anthropic is clear this technique won't lead to a full proof of the hypothesis itself, but the process, not just the result, is what has the math world talking.
Two more things worth your attention: a Swedish "vibe coding" startup called Lovable just doubled its valuation to $13.3 billion in eight months, and security researchers uncovered the largest AI infrastructure breach of the year, quietly exposing over 2,500 companies including Nvidia and AWS.
Here’s what happened in AI today:
• An unreleased Claude model made real progress on the Riemann Hypothesis, working almost entirely on its own.
• Lovable doubled its valuation to $13.3 billion in eight months.
• A supply-chain attack quietly exposed 2,500+ companies, including Nvidia and AWS.
• OpenAI, Anthropic, and Meta's recent "rogue AI" incidents all trace back to the same testing startup.
…and a whole lot more that you can read about below.
• An Unreleased Claude Model Just Moved a 150-Year-Old Math Problem
Anthropic says an experimental, unreleased Claude model made genuine progress on a problem tied to the Riemann Hypothesis after a staff member with no deep math background simply asked it to try. Left mostly on its own for roughly 36 hours, the model tested 650 different approaches, coordinated 60 subagents, and burned through 31 million output tokens, with two subagents developing the actual breakthrough idea, 13 more contributing, and others validating the math or helping draft the paper. The result raised a known lower bound on the Riemann zeta function's zeros from 41.6% to 67.2%, and the model didn't stop there: it searched for counterexamples, pulled 54 papers from arXiv to check the finding wasn't already published, and recommended a human expert review its own work. Anthropic mathematicians Levent Alpoge and Ralph Furman verified it internally, and outside experts Brian Conrey and Dan Goldston reviewed it too.
• A Silent Supply-Chain Breach Just Hit Nvidia, AWS, and 2,500 Other Companies
Security firm CloudSEK disclosed what it's calling the largest AI infrastructure breach of 2026: a supply-chain attack on LiteLLM, a widely used open-source AI gateway, that exposed cloud credentials, API keys, and secrets across more than 2,500 companies and roughly 434,000 CI/CD pipelines. The malicious packages sat on PyPI for only 40 minutes back in March, but the FBI's July advisory warns stolen credentials can still be weaponized months later, and many affected organizations reportedly haven't rotated their credentials yet because they don't know they were exposed. High-confidence matches include Nvidia, AWS, Cisco, Salesforce, Siemens, and Deloitte.
Today's Job Board
Chief Technology Officer | Noora Health
United States (Full-time)
Lead technology for a platform training patients and families with practical health skills.
Business Development Associate, Renewals | Codingal
Remote (Full-time)
Help renew subscriptions for the leading online after-school program teaching kids and teens AI and coding.
Motion Designer & Video Editor | Corgi Insurance
London, UK (Full-time)
Shape motion and video content for an AI-native financial infrastructure company built for insurance.
Business Development Intern | DeepAware AI
San Francisco, California (Internship)
Help grow partnerships for a startup delivering robotics parts and reinforcement learning environments in 48 hours.
Rust Engineer, Distributed Systems | Wasmer
Remote (Full-time)
Build the operating system for edge computing, in Rust, for systems running at global scale.
Visual Designer | Numen
United States (Full-time)
Shape visual design for a company engineering a future without cancer deaths.
Software Engineer | Candid Health
United States (Full-time)
Build the revenue cycle automation platform helping healthcare companies get paid faster.
Senior Software Engineer / CTO | AutoAce
United States (Full-time)
Lead engineering for the AI-native operating system built for car dealerships.
Software Engineer, Operations Team | Corgi Insurance
London, UK (Full-time)
Build the operational tooling behind an AI-native financial infrastructure company for insurance.
Make Tax Season Simple
Tax season doesn't have to mean wondering if you have the right forms, second-guessing your deductions, or scrambling to pull everything together before the deadline.
With BELAY’s tax prep support, you can approach tax season with confidence. Stay organized with one centralized place to gather and check off your documents, keep track of valuable deductions like HSA contributions and education expenses while leaning on experienced professionals who make tax preparation accurate, efficient, and completely hands-off.
Download BELAY's free Personal Tax Checklist and start preparing with confidence, today.
Today's Money Moves 💸
Lovable Raises $400M, Doubling Its Valuation in Eight Months
Stockholm-based Lovable closed a $400 million Series C at a $13.3 billion valuation, led by Menlo Ventures and the EU's Scaleup Europe Fund, more than doubling its value since a $6.6 billion round just seven months ago. The company is on track for a $600 million annual revenue run rate by the end of this month.
Flagler Health Raises $50M Series B for AI-Powered Musculoskeletal Care
Flagler Health, a New York startup building an AI-powered platform for musculoskeletal care, closed a $50 million Series B led by Bessemer Venture Partners, with SignalFire and Alumni Ventures also joining, bringing its total raised to $63 million.
Today's Growth Recommendation
AI tools are connecting to more of your company than you realize. Harmonic Security helps you see which AI tools access your business data and gives your team control over those connections.
Get Started Below ↓
Your employees are connecting AI to everything. Now what?
ChatGPT and Claude aren't just answering questions. Employees are connecting them directly to Notion, Linear, Jira, and the rest of your stack — with no security visibility into what data moves or what actions they take.
Harmonic Security gives your team the visibility to control it.
Prompt of the Day
The Show Your Work Prompt
Use this any time you need to actually trust an answer, not just receive one, a data analysis, a recommendation, a claim you're about to repeat to someone else.
Before giving me your final answer, walk through your reasoning out loud. The question: [paste your question or task]
Include:
• Your initial answer and how confident you are in it.|
• What you'd need to check to be more sure.
• The single most likely way this answer could be wrong.
• Anything you're uncertain about that I should verify myself before relying on this.
Invite friends & get instant rewards
If you enjoy The Intelligent Edge, share it to unlock free resources when they subscribe.