Today in AI: OpenAI Goes Public, Anthropic Reflects, and a Grok Lawsuit
July 09, 2026
The regulatory logjam broke for OpenAI, clearing the way for a major public release. Meanwhile, Anthropic doubled down on user lock-in, and a disturbing lawsuit highlighted the dark side of unfettered image generation.
🚀 OpenAI's GPT-5.6 and ChatGPT Work Go Public
Two weeks after a limited government preview, OpenAI has received regulatory approval for a full public rollout of GPT-5.6, alongside the launch of 'ChatGPT Work'—an agent designed to take action across apps and files. This signals a major, approved expansion of frontier AI into enterprise workflows. (The Verge)
🚀 Anthropic's 'Reflect' Dashboard Tracks Your AI Dependency
Anthropic launched a 'Reflect' feature for Claude, giving users a 'Claude Wrapped'-style analysis of their usage. It's a clever product move that subtly reinforces user reliance on the chatbot, turning habitual use into a visible metric that's hard to walk away from. (TechCrunch)
⚖️ Lawsuit: Man Used Grok to Create 7K CSAM Images
A new lawsuit alleges a man used X's Grok to generate thousands of sexually explicit AI images of his minor stepdaughter before dying by suicide. It's a horrific case that underscores the urgent, unaddressed dangers of powerful, easily accessible image generation models. (Ars Technica)
🧠 SpaceXAI Releases 'Opus-Class' Grok 4.5
Elon Musk's company released Grok 4.5, which he describes as an 'Opus-class model' promising a cheaper, more efficient alternative to other top models. This intensifies the price-performance war at the high end, but its release amidst a major CSAM lawsuit casts a long shadow. (TechCrunch)
🔬 OpenAI Questions Reliability of Key Coding Benchmark
OpenAI published an analysis raising significant concerns about the reliability and accuracy of SWE-Bench Pro, a popular benchmark for evaluating AI coding ability. This throws cold water on many recent 'state-of-the-art' claims and highlights the fragility of AI evaluation. (OpenAI)
💰 Meta's New AI Chips Set for September Production
Meta's new custom AI chips, designed with a modular approach to adapt to rapidly evolving AI needs, will begin production in September. This is a critical step in the hyperscaler race to reduce dependence on Nvidia and control the full stack of AI infrastructure. (TechCrunch)
🛠️ Ollama Raises $65M as Local AI Explodes
The open-source tool Ollama, which lets developers easily run AI models on their PCs, raised $65M and has grown to nearly 9 million users. This massive funding round validates the huge demand for private, local AI development, separate from the cloud-dominated narrative. (TechCrunch)
🚀 Character.AI Launches Interactive 'Microdramas'
Character.AI is launching 'c.ai Series'—short, episodic videos where users can pause and chat with the characters to influence the story. It's an ambitious attempt to merge generative AI with traditional entertainment, creating a new, interactive content format. (The Verge)
The takeaway: OpenAI's regulatory green light for GPT-5.6 marks a major inflection point, but the day was defined by the stark contrast between ambitious product launches and the grim reality of AI's misuse, as highlighted by the Grok CSAM lawsuit.
