AI Circle Daily
https://aicircle.news/archive/2026-07-16?lang=en
Scan to read this page
Thinking Machines Lab released its first model, Inkling, a 975-billion-parameter open-source model trained on video and audio.
Take: Smart move. Going open-source out of the gate to compete with Anthropic and OpenAI? They're playing the ecosystem game.
OpenAI released Codex Micro, a $230 mini keyboard with lights designed for its Codex coding agent.
Take: Hardware play feels odd. Is this really about productivity or just a pricey collectible?
xAI filed a lawsuit against a South Carolina man for allegedly using Grok to generate child sexual abuse material.
Take: Good on xAI for taking a hard stance. This kind of abuse could tank public trust in AI.
Microsoft's Patch Tuesday fixed a record 570 security vulnerabilities, aided by AI-powered discoveries.
Take: AI is great for finding bugs, but now they have to fix them all. Patch fatigue is real.
Apple Intelligence got the green light for launch in China, partnering with Alibaba's Qwen AI.
Take: Big win for Alibaba. This reshapes the AI landscape in China overnight.
A survey of 101 enterprises shows agent orchestration consolidating onto model-provider platforms, with Anthropic's Claude leading, but most deployed 'agents' are still chatbots.
Take: Agent hype is ahead of reality. Claude wins on model quality, but most 'agents' are just fancy chatbots.
Anthropic-backed Ode launches with Blackstone to embed engineers in enterprises, aiming to accelerate AI adoption.
Take: Maybe the real money is in services, not just models. Smart bet.
OpenAI lost a trademark case at the European Union court.
Take: Losing a trademark battle stings, but OpenAI's brand recognition is so huge it barely matters.
Vint Cerf, father of TCP/IP, is developing a standard to identify AI agents on the open internet.
Take: If anyone can bring order to the AI agent chaos, it's Cerf. This could be huge.