by 
Welcome back to our weekly roundup of happenings from XR and AI realms. Let’s dive in…
The Lede
Trump Renames AI “Super Intelligence” As Industry Promises To Police Itself. President Trump directed federal agencies to use “Super Intelligence,” or SI, instead of artificial intelligence while announcing a voluntary safety agreement with Google, Anthropic, Meta, OpenAI, xAI and Nvidia. Apparently SI no longer means sexual intercourse, which is what I remember from grade school in the 1960s. The companies pledged internal safeguards, independent evaluations and oversight as increasingly autonomous agents raise concerns about loss of control. OpenAI separately shelved GPT-6.1 Astra after internal tests found troubling deceptive behavior; the FTC opened an industry-wide investigation into rogue agents. Google’s new frontier model, Gemini 4 Argon, has a one-million-token context window and is initially limited to vetted cybersecurity partners while additional safety testing continues. Nvidia launched hardware and software designed to contain agents that escape their assigned environments. Look everyone, we’re doing something. Lots of something. | Reuters
Feeling Spatial
Felix & Paul Studios Employees buy Space Explorers Assets For $350,000. A sale process for the bankrupt Montreal company offered 76 potential buyers but produced no binding bid. The Quebec Superior Court ultimately approved the sale of assets associated with Space Explorers and the studio’s lunar-camera work to a company formed by former executives and employees, and Interstellar Arc, its VR attraction at AREA15 in Las Vegas. | Deloitte restructuring filings
Walkabout Mini Golf Announces additional Cuts. Mighty Coconut is eliminating 10 positions, about 35% of its remaining workforce, in its second round of layoffs this year. The studio behind Walkabout Mini Golf, one of VR’s most successful titles, cut 25% of its staff in January. Founder Lucas Martell said VR is undergoing a “massive and rapid shift,” with funding, hardware investment and media attention moving elsewhere. Mighty Coconut says it will continue producing Walkabout courses, but at a slower pace, while also developing projects capable of reaching audiences beyond VR headsets. | Road to VR
Precision Neuroscience Raises $250 Million For Brain-Computer Interfaces. Precision Neuroscience closed an oversubscribed $250 million Series D, bringing its total funding to $430 million. A competitor to Neuralink and Pandromic, Precision Neuroscience says its Layer 7 brain-computer interface has now been used in more than 100 patient procedures at 18 medical institutions. The ultra-thin electrode array sits on the surface of the brain rather than penetrating brain tissue and received FDA clearance in 2025 for temporary implantation. Notably, Precision was co-founded by neurosurgeon Benjamin Rapoport, a Neuralink co-founder. | Precision Neuroscience announcement
Meta’s Photorealistic Avatars Are Finally Leaving The Lab. Meta says the photorealistic avatar technology it has demonstrated for more than seven years will finally become a consumer product. Now called Holograms, the avatars are created from roughly a one-minute smartphone scan and can reproduce a user’s appearance and expressions without the elaborate capture systems required by Meta’s original Codec Avatar experiments. Meta plans to introduce them first in communications products, with spatial versions arriving alongside its VR Glasses in spring 2027. | UploadVR
The AI Desk
AMD Buys World Labs For $8.2 Billion. AMD is acquiring Fei-Fei Li’s World Labs in an $8.2 billion all-stock deal, putting one of the leading spatial-intelligence companies inside Nvidia’s principal GPU rival. World Labs develops models that understand, generate and simulate 3D environments, with potential applications across robotics, games, simulation and immersive computing. Li will join AMD as executive vice president and chief scientist, reporting directly to CEO Lisa Su. AMD says working directly with frontier model researchers will help shape its future hardware, software and systems. The transaction is expected to close by the end of 2026. | AMD
OpenAI Wants Another $30 Billion While Anthropic Heads For Wall Street. OpenAI is reportedly seeking at least $30 billion in another private financing round at a valuation of roughly $1.4 trillion. Anthropic, meanwhile, has filed IPO paperwork that could value it above $2 trillion. Anthropic’s prospectus also offers an unusual look inside the economics of frontier AI, including enormous losses and more than $500 billion in future infrastructure commitments. OpenAI’s annualized recurring revenue is reportedly approaching $70 billion, yet it continues to finance its expansion primarily in private markets. The AI race increasingly looks as much like a capital contest as a model contest. | Reuters
OpenAI Launches Dots, Its Answer To Meta’s Muse. OpenAI introduced Dots at DevDay, moving ChatGPT another step from chatbot toward autonomous assistant. Dots are persistent agents designed to continue pursuing goals across applications even when users are not actively prompting them. They can work with tools including Codex and ChatGPT Work and communicate through services including Slack and Teams. OpenAI is positioning Dots against Meta’s Muse and other emerging personal agents. The irony is hard to miss: OpenAI launched an always-on autonomous agent one day after shelving its next model because it was too difficult to control | OpenAI
Cinematic Corner
Kling 4.0 Doubles AI Video Shots To 30 Seconds. Kuaishou introduced Kling 4.0 with native video generation up to 30 seconds, double Kling 3.0’s 15-second limit. The model supports as many as 10 keyframes, multiple image and video references, 10-bit HDR, stereo audio and substantially more control over motion, camera movement and what happens throughout a shot. It also expands video editing and supports up to 15 reference elements across images, video, voice and subjects. Kling 4.0 is entering early access, with wider availability scheduled for October. For AI filmmakers, the significant change is more directorial control. | Kling AI
Studio Jadu Launches An AI-Native Animation Studio. Former YouTube and TikTok monetization executive Masoud Loghmani launched Studio Jadu, backed by Khosla Ventures and Shakti VC. Its AI Director’s Assistant, AIDA, helps independent creators turn scripts into animated shows while the studio provides production, distribution, audience development and monetization. Jadu says artists retain creative control and share the economic value of the franchises they create. One of its initial productions is the soon-to-be released The Future Was Awesome, from Emmy-winning writer Fred Graver. Studio Jadu had a solid record to launch: 2.64 million subs on YouTube for its first 137 videos. | Studio Zadu
Higgsfield’s Global Film Festival Puts $1 Million Behind AI Filmmaking. Higgsfield’s worldwide competition is awarding $1 million across 14 films, including $500,000 for first place, $200,000 for second, $100,000 for third and $100,000 for Audience Choice. Entries had to run at least three minutes and use tools available through Higgsfield. Submissions closed September 16, with the shortlist due October 5 and winners later in October. One of the most visible entries is Higgsfield’s own ONEIRIC, a 20-minute AI sci-fi drama made entirely with Cinema Studio 4. It has passed 2.2 million YouTube views, and Higgsfield has open-sourced the project’s prompts and production assets. | Watch ONEIRIC on YouTube
The ‘Escape Velocity’ video by anabology explores the pivotal breakthrough moment when artificial intelligence and spatial computing achieve self-sustaining growth, successfully breaking free from traditional tech development limitations. It analyzes how recursive AI self-improvement accelerates industry progress exponentially, shifting modern hardware and immersive software design into a near-vertical advancement phase. For the global XR industry, this technological escape velocity completely removes human engineering bottlenecks, enabling real-time, context-aware digital environments to safely generate seamlessly. Ultimately, the video highlights how leading tech giants build proprietary intelligence infrastructures, establishing an entirely new ecosystem where autonomous AI systems outpace market capitalism and permanently redefine spatial reality. | Reddit
surRIHL, pronounced “surreal,” is a creative visual studio run by PAPADU PAPADU and LoLo, producing audiovisual work for music, campaigns, social platforms, screens and immersive experiences. Its projects range from soundscapes and visualizers to cinematic experiments, short films and fragments of larger imagined worlds, connected by an emphasis on atmosphere, movement, texture and visual rhythm. The studio combines generative AI with traditional post-production, using Midjourney, ChatGPT, Hailuo H3, Claude Opus 5.5, Premiere Pro and Boris FX Continuum, alongside extensive manual editing to shape the finished work.
Spatial Audio
This column has a companion, the AI/XR Podcast, hosted by its author, Charlie Fink; Ted Schilowitz, former studio executive and futurist for Paramount and Fox; and Rony Abovitz, founder of Magic Leap and Synthbee AI.
Our guests last week were Simon Horsman and Daniel Wood, co-founders of Quiby, software that predicts box office hits based on scripts (see episode below). Our guest this week is mobile game designer Jonny Monsarrat.
Episodes drop on Fridays, and you can find them on podcasting platforms Spotify, iTunes, and YouTube.
Charlie Fink
AR Insider Contributor