Apple v. OpenAI, and a moving screenless speaker

Apple sues OpenAI over alleged trade secret theft — alleges misconduct "directed by OpenAI's senior leadership, including a long-time former employee." Beyond the poaching drama, this sets a precedent for IP protection in the LLM space — and casts a shadow over any future Apple-OpenAI integration. Two companies that need each other, now lawyering up. — Sarah Perez

RUMOUR

OpenAI’s first hardware device is reportedly a screenless speaker that can move — the bet: a voice-native, mobile smart speaker that skips the screen entirely — and, more pointedly, the smartphone app layer. Low-latency voice plus a hint of embodiment. Whether anyone wants a device that follows them around is another question. — Lucas Ropek

The Eval Flywheel: Turning Every Production AI Failure Into a Regression Test — borrows the oldest trick in software: "That failing test stays in the suite forever, so the bug can never silently come back." The pitch is a pipeline that pulls edge cases from production logs straight into your CI eval suite — an antidote to vibes-based LLM development. — Rashmi

How to Orchestrate 100+ Agents With Claude Code — runs "100+ agents in parallel" through Anthropic's CLI. The real story is orchestration overhead — the bottleneck that shows up long before your model does. A useful stress test of where the tooling actually breaks. — Eivind Kjosbakken

Roblox launches an AI-powered game creation feature in its mobile app — the new "Build" feature lets users generate basic games from a single text prompt, on mobile. A push toward on-device generative 3D that lowers the bar for user-generated content — and takes aim at tools like Hiber3D. — Lauren Forristal

Kimi K3 Intelligence, Performance and Price Analysis — Kimi K3 enters the competitive reasoning model space alongside o1 and DeepSeek-R1; understanding its price-to-performance ratio is vital for global inference strategy. — artificialanalysis.ai