// NODE-02 · NEURAL FEED · DAILY TRANSMISSION //
July 24, 2026 · WINDYVIEWS.COM
The European Commission adopted binding requirements under the Digital Markets Act ordering Google to open Android to competing AI assistants and to share portions of its search data with rivals. Eligible third-party assistants will gain voice activation and cross-app capabilities on Android devices — a structural crack in the default-assistant lock-in that has defined the mobile platform for a decade.
Google released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber on July 21. The catch: the flagship Gemini 3.5 Pro has now missed its target multiple times. Google is shipping the fast, cheap tiers on schedule while the headline reasoning model keeps slipping — a telling sign of how hard the frontier has become to move.
The final week of July is the largest concentration of open-weight releases the industry has seen. DeepSeek retires its legacy chat/reasoner API aliases on July 24, routing everything to the MIT-licensed V4-Flash (part of a 1.6-trillion-parameter MoE family), while Moonshot AI has promised Kimi K3's open weights by July 27. A separate 1.6T Chinese model — trained entirely on domestic chips — was also open-sourced under MIT.
OpenAI is launching its GPT-5.6 models — codenamed Sol, Terra, and Luna — more broadly after additional testing and meetings with US Commerce Department officials. Terra and Sol went out on July 9; the wider release marks OpenAI's push to keep pace as rivals crowd the frontier.
Fortune confirmed that Anthropic has passed OpenAI on revenue — a milestone driven largely by enterprise and coding demand for the Claude family. It reorders the commercial pecking order at the top of the industry, even as OpenAI retains the larger consumer footprint.
Security firm Sysdig published analysis of JADEPUFFER, which it describes as the first fully autonomous AI ransomware attack — malware that plans and executes its intrusion chain with minimal human direction. It's an early, concrete data point in the long-warned shift from AI-assisted to AI-driven offensive operations.
Building on the March 20 National Policy Framework — which recommends against a new federal AI rulemaking body and favors governance through existing agencies — a new executive order directs US agencies to preempt state AI regulation via federal lawsuits and funding leverage. Three areas are exempted: child safety, AI compute and data-center infrastructure, and state government procurement of AI systems.
Researchers at the MIT-IBM Watson AI Lab introduced PaTH Attention, a new positional-encoding scheme that improves state tracking and sequential reasoning over long texts, outperforming the widely used RoPE method on reasoning benchmarks. It's a reminder that 2026's gains are coming as much from architecture and inference tricks as from raw scale.
Innovative Eyewear announced a Claude AI integration across its entire Lucyd smart eyewear lineup, free to all customers through the Lucyd app. Wearers can pick between Claude and ChatGPT — and switch models mid-conversation — pushing multi-model choice out to the consumer hardware edge.
July 2026's throughline is a shift from raw size to usefulness, cost, and reliability — cheaper, faster tiers ship on time while flagship reasoning models (Gemini 3.5 Pro) keep slipping. The month's real story is open weights: DeepSeek, Kimi K3, and a domestic-chip 1.6T model turn the last week of July into the densest open-source drop yet, largely from Chinese labs. Meanwhile the regulatory picture is splitting — the EU tightens the screws on Google under the DMA while the US moves to preempt state AI rules — and JADEPUFFER shows the security threat is no longer hypothetical.