🔥 The Week in AI News - July 13 - 17, 2026

Learn more about my services: www.promethean-ai.com

Google, OpenAI, Roblox, Spotify, Meta, Instagram, Instacart, Canva, YouTube, DoorDash, Moonshot, Anthropic, Apple, Alibaba, Baidu, Thinking Machines, Suno, Deezer, Genius, Pinterest and more.

▪️ Google Vids turned users into the stars of their own AI-generated videos.

Google added personal avatars that could replicate a user’s appearance and voice from a selfie and short recording.

Gemini Omni also brought reference-image generation, background replacement, lighting fixes, effects, and conversational editing into the video platform.

▪️ OpenAI’s first major consumer device reportedly took shape as a screenless AI companion for the home.

The device was described as a mobile smart speaker capable of learning about its owner, accessing connected services, and moving through built-in mechanical components.

OpenAI appeared to be aiming beyond the traditional voice assistant toward a physical version of ChatGPT with a persistent personality.

▪️ Google renamed NotebookLM and pulled it deeper into the Gemini ecosystem.

NotebookLM became Gemini Notebook while remaining available as a standalone application.

Google also expanded its connections to Gemini and Search, added plans for notebook access inside AI Mode, and began rolling out secure cloud-based coding capabilities.

▪️ OpenAI launched a $230 keyboard that turned managing AI agents into a physical workflow.

The Codex Micro used illuminated agent keys, programmable controls, a joystick, and a dial for adjusting how much reasoning an agent applied to a task.

OpenAI also introduced ChatGPT Work as a broader command center for managing long-running AI work across coding, research, finance, engineering, and data analysis.

▪️ Roblox let anyone build a mobile game by describing it to AI.

Its new Build feature transformed simple text prompts into playable experiences complete with environments, characters, mechanics, sound, and visual styling.

Roblox said its recommendation system would prioritize games that retained real players rather than flooding discovery with disposable AI-generated content.

▪️ Spotify turned music discovery into an ongoing conversation.

Premium users gained the ability to speak or type requests, refine recommendations through follow-up questions, and ask about their listening histories.

The assistant could also save tracks, manage queues, follow artists, and help users explore podcasts, audiobooks, genres, and unfamiliar music.

▪️ New workforce research warned that America was heading toward its largest labor shortage ever.

The projected gap was driven by mass retirements, lower birth rates, reduced immigration, declining college enrollment, and millions of people leaving the workforce.

At the same time, employers continued struggling to fill essential roles while graduates earned degrees that did not consistently match the skills the economy needed.

▪️ Meta removed an Instagram AI feature after users revolted over how easily it could reuse their images.

The tool had allowed people to generate new content by referencing photos from public Instagram accounts without notifying the original users.

Meta admitted the feature had missed the mark and withdrew it following criticism from users, creators, and talent agencies.

▪️ Google connected AI Mode directly to the apps people already used.

Users could link services including Instacart, Canva, and YouTube Music to complete tasks without leaving conversational Search.

Google effectively moved AI Mode from providing recommendations to adding groceries, finding templates, building playlists, and initiating real actions across outside platforms.

▪️ DoorDash opened its ordering system to AI agents through the command line.

The limited DoorDash CLI beta allowed developers and their agents to search stores, compare deals, and complete purchases without opening the consumer app.

What looked like a developer joke was actually an early example of agentic commerce becoming a programmable infrastructure layer.

▪️ OpenAI sold a ChatGPT basketball because apparently AI needed branded outdoor equipment too.

The $70 rubber basketball accompanied a broader merchandise campaign encouraging users to step away from their screens and make room for physical creativity.

It was less a serious AI product than a strange signal that OpenAI increasingly saw itself as a consumer lifestyle brand.

▪️ Moonshot prepared Kimi K3 to challenge the most powerful closed AI models.

The upcoming open-weight model was reportedly expected to match or surpass Anthropic’s Opus 4.8 in some areas while reaching between two and three trillion parameters.

Its arrival could intensify pressure on companies charging premium prices for proprietary frontier models.

▪️ Apple secured approval to bring Apple Intelligence into China with help from Alibaba and Baidu.

Alibaba’s Qwen models were set to support text and image intelligence across Apple’s operating systems, while Baidu confirmed it was also developing features for Chinese users.

The partnerships gave Apple a long-delayed path into one of its most important markets without relying entirely on its own models.

▪️ Thinking Machines released Inkling as an open challenge to one-size-fits-all AI.

The company’s first in-house model used a mixture-of-experts architecture and was designed for organizations that wanted to download, modify, and fine-tune their own intelligence.

Thinking Machines argued that adaptable models shaped around proprietary expertise could outperform larger general-purpose systems at a fraction of the operating cost.

▪️ A hack reportedly exposed how Suno may have collected enormous amounts of copyrighted training data.

Source code accessed during the breach allegedly showed systems for scraping audio from YouTube Music, Deezer, Genius, stock libraries, and podcast feeds.

The claims added new pressure to Suno’s fair-use defense and raised separate concerns about customer information reportedly exposed during the intrusion.

▪️ Apple opened its rebuilt Siri to the public after years of delays and false starts.

The iOS 27 public beta gave everyday users access to a Siri that could understand on-screen activity, search personal information, use broader world knowledge, and operate more deeply across Apple devices.

The release represented Apple’s largest real-world test yet of whether its assistant could finally compete with ChatGPT, Gemini, and Claude.

▪️ Google redesigned Images as a personalized discovery feed built to compete with Pinterest.

The new experience introduced an endless “For You” gallery, interest-based recommendations, saved collections, and continuously refreshed visual inspiration.

Google also brought AI image generation into Search, allowing users to create visuals when the image they wanted did not already exist online.

▪️ New York became the first state to halt permits for major new data centers.

The temporary order applied to projects using at least 50 megawatts while the state developed a new environmental review process.

Growing concern over electricity prices, grid capacity, water consumption, noise, and local control had officially begun colliding with the AI industry’s infrastructure race.

▪️ The European Union ordered Google to loosen its grip on Android and Search.

Google was required to give competing AI assistants comparable access to Android hardware, apps, voice controls, and user-approved data.

The company also had to share certain Search data with rival search engines and AI services, potentially allowing tools such as ChatGPT and Claude to compete more directly with Gemini across Google’s own platforms.