Apple logs every tap, Meta guts engineering, OpenAI loses $38.5B

Apple’s new Personalized Collections feature records every tap you make in the App Store and sends the data unencrypted, tied to your Apple ID. The capture includes typing speed and cannot be disabled, raising a major privacy red flag for users and developers alike. It gives Apple deeper insight for ad‑targeted recommendations.

A recent blog warns that treating large language models as the core executor of business processes is a structural flaw. It argues LLMs excel at translating messy inputs, while deterministic rules, audits, and cost‑sensitive logic belong in specialized, observable systems. The shift protects reliability and governance in mission‑critical software.

China’s universities have eliminated 12,200 undergraduate programs, including product design, to align with a national AI push. The purge reshapes the talent pipeline, forcing designers to blend AI fluency with creative skills or risk obsolescence in a fast‑changing job market.

A2UI is an open‑source spec where agents send a declarative JSON tree describing UI components. The client renders these with its native toolkit, keeping code safe while supporting incremental updates across Flutter, React, SwiftUI, and more. Designers can now pre‑define component catalogs and let agents dynamically compose rich, cross‑platform interfaces.

Leaked audited documents reveal OpenAI spent $34 billion in 2025 while generating $13.07 billion in revenue, resulting in an operating loss of about $38.5 billion. The scale of the burn forces the company to lean heavily on investors and shapes its upcoming IPO strategy.

Meta’s leadership has launched an AI‑obsessed overhaul that slashes engineering headcount and reclassifies the org from profit‑center to cost‑center. The shift dismantles the long‑standing “move‑fast” culture, risking talent loss and product instability. It signals a broader reckoning for how big tech balances AI ambition with sustainable engineering.

SpaceX agreed to buy AI code editor Cursor for $60 billion, thrusting Elon Musk into the AI developer‑tools arena. The deal intersects with Anthropic’s contested Fable system and the persistent AI jailbreak problem, suggesting a tighter tie between aerospace ambitions and AI coding capabilities. Expect faster integration of AI‑generated code into SpaceX projects and heightened regulatory scrutiny.

Rocketgraph slashes billions of log lines into a handful of structural templates, then spots anomalies in seconds. The engine runs locally, never sends data to SaaS, and can hand a concise snapshot to an LLM for AI‑driven debugging without losing signal.

SpaceX agreed to acquire Cursor, the AI‑powered coding tool behind Anysphere, in an all‑stock deal worth $60 billion. The move gives SpaceX’s xAI division a commercial‑grade product and a foothold in the fast‑growing enterprise AI tools market, a key revenue source the company promised investors.

Freebuff is a free CLI AI coding assistant that runs in the terminal and is funded by ads instead of subscriptions. It offers 5‑10× speed boosts with fast models, built‑in web research, and optional ChatGPT integration for deeper planning.

Lenny breaks down four loop types, heartbeat, cron, hook, and goal, and the five components any production‑ready loop needs. He then builds a daily PR‑review loop in Claude Code and a weekly skills‑identification loop in Codex that run without human supervision, cutting token waste and eliminating manual PR babysitting.

Reyn is a Mac‑only AI that watches your screen, logs activity locally, and lets you query past work without ever sending data to the cloud. It adds daily briefs, shareable workflows, and multi‑model support, giving privacy‑first professionals instant recall of their own digital history.

Draft runs a background daemon that captures context from Slack, GitHub and meetings, then injects it into Claude‑based agents at start of each session. Teams get seamless continuity, eliminating the need to re‑brief AI every time they switch tasks. Currently macOS‑only, but open‑source and extensible.

Genesis AI unveiled Eno, a minimalist, wheeled robot that combines its GENE AI brain with dexterous hands to reason, adapt, and own outcomes beyond preset tasks. Backed by ex‑Google CEO Eric Schmidt and partnered with LG, Eno targets factories, labs, and eventually consumer spaces, promising a new scale of human‑machine collaboration.