Issue catalog
AI Breakthroughs in Math, Cursor's $3B Milestone, and Spotify's AI Expansion
AIDaily issue

AI Breakthroughs in Math, Cursor's $3B Milestone, and Spotify's AI Expansion

Google DeepMind solves nine open math problems, Cursor hits $3B in revenue with a potential SpaceX acquisition, and Spotify launches new AI-driven audio tools.

Podcast В· 2 min

AI Breakthroughs in Math, Cursor's $3B Milestone, and Spotify's AI Expansion

Google DeepMind solves nine open math problems, Cursor hits $3B in revenue with a potential SpaceX acquisition, and Spotify launches new AI-driven audio tools.

Research & Breakthroughs

Google DeepMind's AlphaProof Nexus Solves Nine Erdős Problems

Google DeepMind has introduced AlphaProof Nexus, an AI system capable of generating machine-verified mathematical proofs. The system successfully solved nine open Erdős problems—some of the most challenging unsolved questions in mathematics—including two that had remained unsolved for 56 years. The process involved pairing an LLM with Lean, a proof assistant, to iteratively generate and verify proofs. Each problem was solved at a cost of a few hundred dollars, and the system also proved 44 open conjectures from the Online Encyclopedia of Integer Sequences.

This development follows recent progress by other labs in applying AI to complex scientific challenges. By automating the verification process, Google's approach demonstrates how AI can move beyond simple pattern recognition to contribute to original mathematical discovery. The ability to generate and verify proofs at machine speed suggests a shift in how researchers may approach long-standing theoretical problems in the future.

Anthropic's Claude Mythos Identifies 10,000+ Vulnerabilities

Anthropic has released initial results from Project Glasswing, revealing that its restricted cybersecurity model, Claude Mythos, identified over 10,000 high- or critical-severity vulnerabilities in software projects within one month. Partners, including Cloudflare and Mozilla, utilized the model to detect and patch bugs, with Cloudflare reporting a false positive rate superior to human testers. Additionally, the model was used by a partner bank to detect and block a $1.5 million fraudulent wire transfer.

Anthropic continues to gate Mythos, citing the lack of sufficient safeguards to prevent potential misuse. However, the success of the model highlights the growing capability of AI in automated security auditing. As frontier models become increasingly adept at identifying complex software flaws, the industry faces a race between AI-driven defense and the potential for automated exploitation.

Business & Industry

Cursor Hits $3B Annual Revenue Amid SpaceX Acquisition Talks

AI-powered coding editor Cursor has reported $3 billion in annualized revenue, marking a significant growth trajectory from $2 billion in February. The company is projecting revenue to exceed $6 billion by the end of 2026. This rapid scaling has drawn interest from SpaceX, which reportedly secured an option to acquire the startup for $60 billion, or pay a $10 billion walk-away fee. The deal follows a period of close collaboration, including Cursor's use of SpaceX's Colossus supercomputer for model training.

Cursor's success underscores the increasing value of the 'application layer' in the AI stack—tools that integrate frontier models into specific, high-value workflows like software development. The potential acquisition by SpaceX suggests a strategic move to secure proprietary coding infrastructure, reflecting a broader trend where major tech players are integrating specialized AI dev-tools to maintain competitive advantages.

DeepSeek Secures $10B Funding and Cuts Prices

DeepSeek has announced a $10 billion funding round while simultaneously slashing prices for its V4-Pro model by 75%. The new pricing brings costs down to $0.435 per million input tokens and $0.87 per million output tokens, significantly undercutting many closed-source competitors. The lab's founder, Liang Wenfeng, has stated that the primary goal of the organization is the achievement of AGI, and the company plans to continue its strategy of releasing open-source models despite the massive capital raise.

This aggressive pricing strategy puts further pressure on the AI market, forcing competitors to justify the premium costs of their proprietary models. By combining high-level funding with a commitment to open-source releases and low-cost access, DeepSeek is positioning itself as a major disruptor in the race to develop and deploy frontier-capable AI.

Spotify Launches AI Remixing and Audio Tools

Spotify has announced a suite of new AI-driven features following its annual investor day. A landmark deal with Universal Music Group (UMG) will allow fans to create AI-generated covers and remixes of UMG artists' songs, with a framework for artist consent, credit, and revenue sharing. Additionally, the company introduced 'Studio by Spotify Labs,' a desktop app that converts notes and emails into personalized audio briefings, and an ElevenLabs-powered tool that enables authors to create audiobooks without professional recording equipment.

These moves represent Spotify's strategy to expand beyond music streaming into broader audio creation and consumption. By formalizing a licensing framework for AI-generated fan content, Spotify is attempting to address the tension between AI innovation and artist rights, potentially setting a standard for the music industry's approach to generative AI.

Policy & Retail

Starbucks Retires AI Inventory Tool

Starbucks has officially retired its AI-powered inventory management tool across its 11,000+ North American locations after nine months of operation. The system, which was intended to automate inventory counting, reportedly struggled with accuracy, including difficulties distinguishing between different types of milk. The company has reverted to manual counting methods.

This incident serves as a notable example of the challenges in deploying AI in physical retail environments. While AI holds promise for operational efficiency, the failure of the tool highlights the gap between theoretical model performance and the complexities of real-world, high-volume retail operations.