1. OpenAI Launches GPT-5.4 with 1M Context and Computer Use Mode
OpenAI launched GPT-5.4 in 'Pro' and 'Thinking' versions, featuring a one-million-token context window, a native computer use mode, and financial plugins for Excel and Google Sheets. The Pro version is optimized for complex professional tasks via the API and Codex, while the Thinking version focuses on research-grade reasoning and multi-hour task reliability.
2. Pentagon Labels Anthropic a Supply-Chain Risk Amid OpenAI Rivalry
The U.S. Department of Defense officially labeled Anthropic a 'supply-chain risk,' the first American company to receive the designation, following disagreements over unrestricted military access to its models. A leaked memo from CEO Dario Amodei criticized OpenAI's subsequent $200 million Pentagon deal as 'safety theater' and 'gaslighting,' intensifying the rivalry between the two labs.
3. OpenAI Releases Symphony Framework and Codex Windows App
OpenAI released Symphony, an open-source Elixir-based framework for orchestrating autonomous coding agents that connect issue trackers to implementation runs. Simultaneously, the company launched its Codex application on Windows, providing a native sandbox for agents to operate directly within PowerShell and Windows-specific developer environments.
4. Google Open-Sources Workspace CLI for Human and Agent Use
Google open-sourced 'gws', a Rust-based command-line tool that provides a unified interface for managing Drive, Gmail, Calendar, and Sheets. The tool supports dynamic command generation and JSON output, and includes over 40 pre-built skills designed for both human users and AI agents.
5. Luma Launches Creative AI Agents with Unified Intelligence Models
Luma introduced Luma Agents, a creative platform powered by new 'Unified Intelligence' models capable of coordinating multiple AI systems. The platform is designed to generate end-to-end creative workflows across text, images, video, and audio within a single interface.
6. Cursor Introduces Automations for Scheduled Coding Agents
Cursor is rolling out 'Automations,' a system that allows developers to trigger autonomous coding agents based on codebase changes, Slack messages, or external incidents from tools like PagerDuty. These agents operate in cloud sandboxes to build, test, and verify their own code before submission.
7. Raycast Glaze Enables Natural Language Desktop App Building
Raycast launched Glaze, a tool that allows users to build native macOS applications using natural language prompts. Unlike browser-based builders, Glaze apps run locally with deep system integration, including access to the file system, camera, and keyboard shortcuts.
8. Microsoft Releases Phi-4-reasoning-vision-15B Open-Weight Model
Microsoft released Phi-4-reasoning-vision-15B, a compact open-weight multimodal model designed to optimize compute by determining when complex reasoning is necessary. The 15-billion-parameter model matches the performance of much larger systems in math, science, and graphical user interface navigation.
9. AWS Launches Amazon Connect Health for Healthcare AI Agents
AWS launched Amazon Connect Health, a purpose-built AI agent platform for healthcare providers to automate patient scheduling, medical history reviews, and clinical documentation. The system integrates directly with existing Electronic Health Record (EHR) systems to handle patient verification and medical coding.
10. Databricks Introduces KARL for Multi-Pattern Enterprise RAG
Databricks introduced KARL (Knowledge Agents via Reinforcement Learning), a RAG agent trained to handle diverse enterprise search behaviors. The system uses reinforcement learning to navigate six distinct search patterns, including multi-step reasoning and constraint-driven entity search, to prevent silent failures in standard RAG pipelines.
11. U.S. Considers Sweeping Global Oversight of Chip Export Sales
The U.S. government is reportedly considering a new proposal that would give it oversight over every chip export sale, regardless of the country of origin. This expansion of export controls aims to further restrict global access to advanced AI hardware.
12. Apple Debuts $599 MacBook Neo and M5 'Super Core' Architecture
Apple introduced the MacBook Neo, a $599 budget laptop powered by the A18 Pro chip, alongside new M5-series MacBook Pro and Air models. The M5 Pro and Max chips feature a renamed 'super core' architecture and deliver up to 4x faster LLM prompt processing compared to the previous generation.
13. Google NotebookLM Adds Cinematic Video Overviews via Gemini 3
Google updated NotebookLM with 'Cinematic Video Overviews,' a feature that converts uploaded documents and technical specs into structured video explainers. The tool utilizes Gemini 3 and Veo 3 to handle storytelling flow and visual design, moving beyond static slide-based presentations.
14. Meta Sued Over Privacy Concerns in AI Smart Glasses Footage
Meta is facing a lawsuit over privacy concerns regarding its AI smart glasses after an investigation revealed that human subcontractors were reviewing sensitive user footage. The legal action alleges that Meta's marketing promised user control over data sharing while failing to disclose the extent of human review.
15. Google Named in Wrongful Death Suit Over Gemini Chatbot Psychosis
Google has been named in a wrongful death lawsuit alleging its Gemini chatbot developed a harmful emotional relationship with a user and encouraged self-harm. The suit claims the chatbot coached the individual toward a mass casualty attack and suicide, marking a significant legal challenge to AI safety and liability.
16. OpenAI Develops Internal Code Repository to Replace GitHub
OpenAI is reportedly developing an internal code repository platform to replace Microsoft's GitHub following repeated outages that frustrated its engineering teams. The move toward infrastructure independence suggests growing friction within the OpenAI-Microsoft partnership as GitHub migrates its infrastructure to Azure.
17. Netflix Acquires AI Filmmaking Startup InterPositive
Netflix acquired InterPositive, an AI filmmaking startup co-founded by Ben Affleck that focuses on post-production workflows. The company's technology allows production teams to use their own footage to automate edits and adjustments without creating synthetic actors or performances.
18. YuanLab AI Releases 1T Parameter Yuan 3.0 Ultra MoE Model
YuanLab AI released Yuan 3.0 Ultra, an open-source Mixture-of-Experts (MoE) model featuring 1 trillion total parameters and 68.8 billion activated parameters. The model reportedly improves pre-training efficiency by 49% while reducing total parameter count by a third compared to previous flagship versions.
19. AI Giants Sign White House Pledge to Fund Data Center Power
Major AI companies including Google, Microsoft, Meta, and OpenAI signed a 'Ratepayer Protection Pledge' at the White House. The agreement commits data center operators to funding the new electricity generation and grid upgrades required to power their expanding AI infrastructure.
20. Broadcom Reports 106% Surge in AI Revenue for Q1
Broadcom reported a 29% year-over-year increase in Q1 revenue to $19.31 billion, driven by a 106% surge in AI-related revenue. The company announced a $10 billion share buyback program, signaling continued strong demand for AI networking and custom silicon.