Blockade Labs just upgraded its Skybox AI platform with Model 3, enabling creators to generate and download seamless 360-degree environments in 8K resolution. The new model packs 77% more pixels to target Apple Vision Pro headsets, alongside a new remix tool for existing panoramic images.
Meta just launched Llama 3, delivering 8B and 70B open-access models trained on over 15 trillion tokens. The smaller 8B variant can easily run and fine-tune on consumer GPUs down to 4-bit precision, backed by an expanded 128k vocabulary. Why it matters: Meta is resetting the benchmark for open AI, giving developers frontier-level capabilities without proprietary API lock-in.
NVIDIA just rolled out full-stack optimizations for Meta's brand-new Llama 3 across its data center GPUs, edge devices, and PCs. In early benchmarks, a single H200 GPU churned out roughly 3,000 tokens per second on the 70B model, enough to handle 300 users simultaneously. The 8B model is also landing on RTX PCs and Jetson Orin modules.
Meta just upgraded its Meta AI assistant with Llama 3 and launched a standalone web experience at meta.ai. The assistant is woven directly into the search bars of WhatsApp, Instagram, Facebook, and Messenger, rolling out across more than a dozen English-speaking countries with real-time text-to-image generation as you type.
DARPA just revealed that an AI-controlled fighter jet successfully went head-to-head against a human pilot in an in-air dogfight. Flying the experimental X-62A against a human-piloted F-16, the two jets reached speeds of 1,200 miles per hour and got within 2,000 feet, with human safety pilots never needing to take back control.
Nothing just refreshed its earbud lineup with the Nothing Ear and Ear (a), adding built-in ChatGPT access triggered by a pinch of the stem. Instead of inventing dedicated AI gadgets from scratch, Nothing skipped a step by baking conversational AI into everyday earbuds—provided you also carry a Nothing phone.
HarperCollins just teamed up with ElevenLabs to turn select non-English backlist series into audiobooks using AI text-to-speech. Traditional studio recording has historically been far too expensive for older, niche titles, but ElevenLabs' Studio tool can produce a full audiobook in as little as an hour. While human narrators will still handle primary releases, AI is quietly stepping in to revive forgotten catalogs abroad.
Intel just unveiled Hala Point, the world’s largest neuromorphic computer packing 1.15 billion artificial neurons across 1,152 Loihi 2 chips into a microwave-sized chassis. By storing and processing data simultaneously like a real brain, it solves optimization problems with 100 times less energy than conventional machines. While not built to train LLMs from scratch, it opens a promising path for ultra-efficient continual learning once software catches up.
Microsoft just rolled out industrial AI solutions designed to bridge factory-floor data with generative tools, including a factory operations Copilot template on Azure AI and data integration in Microsoft Fabric. Front-line workers can use natural language to troubleshoot malfunctioning machinery and query operations in real time. Core features are currently in private preview.
Amazon just detailed the roadmap for its checkout-free tech, planning to more than double third-party Just Walk Out deployments in smaller stores while pushing smart Dash Carts in larger supermarkets. The company also pushed back on reports that Indian reviewers manually watched video feeds to tally bills, clarifying that computer vision models handle checkout automatically while humans only annotate training data. Why it matters: Amazon isn't giving up on cashierless retail—it's tailoring sensor fusion to quick-trip convenience and saving smart carts for big grocery hauls.
Stability AI just launched Stable Diffusion 3 and its Turbo version on its Developer Platform API, partnering with Fireworks AI for enterprise-grade availability. Powered by the new MMDiT architecture, the model rivals DALL-E 3 and Midjourney v6 in typography and prompt adherence. Model weights will also be available soon for member self-hosting.
Boston Dynamics just introduced an all-electric Atlas humanoid, ditching its iconic hydraulics for custom high-powered electric actuators. The new design is sleeker, lacks exposed cables, and can pull off uncanny 180-degree joint rotations to stand straight up from the floor. Commercial pilot testing kicks off at Hyundai facilities early next year.
Alphabet’s X lab just teamed up with the U.S. National Guard to speed up emergency response with AI. Built by its Bellwether team, the system analyzes aerial photos of disaster zones in seconds to pinpoint damaged critical infrastructure, replacing a slow manual triage process ahead of this June's wildfire season.
Logitech just launched the Logi AI Prompt Builder inside its Logi Options+ app, alongside a new $49.99 Signature AI Edition mouse with a dedicated AI trigger. The tool automatically captures highlighted text and lets you run preset or custom prompts in seconds. It basically turns daily LLM queries into a single hardware click.
Google DeepMind CEO Demis Hassabis just revealed that Google will eventually invest more than $100 billion into AI over time, topping the rumored cost of Microsoft and OpenAI's Stargate supercomputer. Hassabis insisted Google's processing power already outstrips rivals, with capital expected to flow into Axion chips and massive data center buildouts.
Quora's Poe just rolled out multi-bot chat, letting users summon multiple AI models inside the same thread. You can simply @-mention bots like GPT-4, Claude 3, or Gemini to compare answers side-by-side or hand off tasks across them in a single workflow. It basically turns chaotic tab-switching into an organized relay team.
Amazon Music just rolled out Maestro, a new AI tool in beta that turns text prompts, voice notes, and even random emojis into custom playlists. Currently testing among select U.S. mobile users, Unlimited subscribers can stream tracks instantly, while free and Prime listeners are limited to 30-second previews. It's Amazon's direct answer to Spotify's recent AI push, designed to nudge more free users into paid tiers.
Apple is reportedly planning to run its initial batch of iOS 18 AI features entirely on-device without cloud servers. Basic capabilities like smart text generation in Messages, Safari, and Siri will work offline to ensure privacy, while heavier tools may tap cloud partners. Apple is expected to detail the plans at WWDC on June 10.
Adobe just previewed new generative AI tools coming to Premiere Pro later this year, letting editors add or remove objects from shots and extend short clips with AI-generated frames. It also teased integrations with third-party models like OpenAI's Sora and Pika to automatically generate B-roll directly in the timeline. Why it matters: professional editors might soon get text-to-video superpowers baked right into their standard workflow.
Blackmagic Design just launched the public beta of DaVinci Resolve 19, packing a slew of AI-powered creative features. Editors can now trim clips directly by editing transcribed text, track and stabilize footage using the neural-engine IntelliTrack, and isolate audio with Dialogue Separator FX. The standard version remains completely free, while the Studio upgrade sits at $295.
PaintCam just launched Eve, an AI-powered home security camera that actively fires paintballs and tear gas at intruders who ignore its warnings. The system uses computer vision to distinguish friends from foes, but turning backyard defense into autonomous armed deterrence opens up a minefield of legal and ethical questions.
Hugging Face just released Idefics2, an open-source 8B vision-language model licensed under Apache 2.0. It preserves native image resolutions, significantly upgrades OCR, and punches way above its weight class on visual QA benchmarks against 30B+ rivals. Why it matters: Developers get an efficient multimodal base model ready to fine-tune directly in Transformers.
Marques Brownlee (MKBHD) just dropped his review of Humane's AI Pin, calling it “the worst product I’ve ever reviewed.” After testing it for a week, Brownlee found the wearable slow to respond, barely visible outdoors, and plagued by poor battery life. Why it matters: It delivers a brutal reality check to first-gen AI hardware, warning buyers never to purchase a device based on future software promises.
OpenAI just launched its first Asian office in Tokyo and rolled out an early-access GPT-4 model specifically optimized for the Japanese language. The custom build runs up to 3x faster with significantly lower token costs for translation and summarization, alongside plans for a wider API rollout in the coming months.
Meta is now testing its Meta AI assistant directly inside Instagram's search bar. Queries can kick off a private DM conversation for questions or surface specific Reels using pre-loaded prompts, mimicking Perplexity's clean interface. It looks like Meta's latest push to fix Instagram's notoriously clunky search and better take on TikTok.
Grok-1.5V just dropped as a first-generation multimodal model, expanding beyond text to handle documents, diagrams, and real-world photos. The team also released RealWorldQA, a new spatial reasoning benchmark where Grok scored 68.7%, edging out rival frontier models. It will roll out soon to early testers and existing users.
SAG-AFTRA just struck a tentative five-year deal with Warner Music, Sony Music, and other major labels to guard against AI voice cloning. Under the agreement, labels must secure consent and pay compensation before dropping any track that uses a digital replica of an artist's voice—and "singer" is legally reserved for humans.
Freepik just launched an updated AI Image Generator featuring real-time image rendering and an infinite scroll workflow. Users can generate photorealistic variations in a single click across multiple styles. Free accounts get 20 images daily, while premium users unlock unlimited access.
Dyson just announced CleanTrace, an AR feature that turns vacuuming into a video game by digitally painting over areas you've already cleaned on your phone screen. It relies on the lidar sensor on an iPhone Pro mounted to your vacuum handle. The catch? It only works with the $1,000 Gen5detect and needs a dedicated clamp.
Google just announced that starting May 15, its Magic Editor tools in Google Photos will be free for all iOS and Android users. That includes features like Photo Unblur and Magic Eraser, though non-Pixel users will be capped at 10 free saves per month unless they pay for Google One.
The Verge just dropped its review of the Humane AI Pin, and the verdict is brutal: it's not even close to replacing your phone. Despite costing $699 plus a $24 monthly fee, the screenless wearable constantly overheats, dies in just a few hours, and can’t even set basic alarms or manage calendar events. Why it matters: While the post-smartphone dream is alluring, current hardware and AI models clearly aren't ready to handle ambient computing on their own.
Adobe is paying creators an average of $2.62 to $7.25 per minute for short clips of everyday human actions and emotions to train its upcoming text-to-video generator. The data-buying push is aimed directly at countering OpenAI’s Sora, doubling down on Adobe’s pitch of commercially clean, copyright-safe training data.
HubSpot just teamed up with AI creators Matt Wolfe and Nathan Lands to launch The Next Wave, a weekly podcast breaking down practical AI strategies in 40 minutes or less. The debut episode features Perplexity CEO Aravind Srinivas discussing the future of search and open-source models. It is built to serve as an on-demand Chief AI Officer for growing companies.
Udio just opened its AI music creation app to the public, letting anyone generate fully mastered tracks with vocals in under 40 seconds from text prompts. Backed by a16z alongside music icons like will.i.am, the platform also features text-based remixing and song extension. Why it matters: Udio brings high-fidelity music production to anyone with an idea, intensifying competition in generative audio.
Intel just unveiled its Gaudi 3 AI accelerator at Vision 2024, claiming it outperforms Nvidia's dominant H100 GPU. Built on TSMC's 5nm process with 128GB of HBM2e memory, Intel says Gaudi 3 runs models like LLaMA up to 1.7x faster and is 2.3x more power-efficient. Air-cooled models roll out in Q3, followed by liquid-cooled units in Q4.
Meta chief AI scientist Yann LeCun just poured cold water on AGI hype, arguing that large language models will never reach human-level intelligence. He claims LLMs lack fundamental traits like reasoning, planning, and physical world awareness, suggesting the industry must pivot toward sensor- and video-trained "world models" instead.
Meta just unveiled the next generation of its custom AI inference accelerator, MTIA. Built on TSMC's 5nm process, the new silicon more than doubles compute and memory bandwidth, delivering 3x the performance of MTIA v1 across key models. Why it matters: Meta is scaling its in-house chip lineup to slash inference costs and ease reliance on commercial GPUs.
Mistral AI just dropped its largest open-source model yet, releasing Mixtral 8x22B via an unceremonious torrent link on X. The sparse MoE model features 176 billion total parameters with 44 billion active per token and a 65,000-token context window. Released under a permissive Apache 2.0 license, it delivers massive compute efficiency to open-source builders without corporate guardrails.
US Representative Adam Schiff just introduced the Generative AI Copyright Disclosure Act, a bill that would force companies to reveal copyrighted material used in their models. AI makers would have to file a record of protected works with the Register of Copyrights at least 30 days before releasing new tools, or face financial penalties. It wouldn't ban using copyrighted data, but it aims to crack open the secretive black box behind systems like ChatGPT.
OpenAI just dropped a majorly improved GPT-4 Turbo model in its API and began rolling it out to ChatGPT. GPT-4 Turbo with Vision is now generally available, finally letting developers use JSON mode and function calling directly on visual inputs. Why it matters: Multimodal workflows can now return reliable, structured data straight from images.
Meta just confirmed that its next-generation open-source model, Llama 3, will begin rolling out within the next month. The company plans to release multiple versions across the year and integrate them throughout its product lineup. Llama 3 aims to relax previous over-cautious restrictions, with reports expecting parameter sizes to reach roughly 140B.
Google just released CodeGemma in collaboration with Hugging Face, bringing a family of open-access code specialists built on Gemma. Trained on an extra 500 billion tokens, the lineup includes an infilling-focused 2B model that runs on consumer hardware alongside 7B base and instruct variants. Why it matters: It offers developers fast, private code completion without needing massive cloud infrastructure.
Apple just made a big push to bring Vision Pro into corporate workspaces. Powered by on-device ML and M2's Neural Engine, the headset integrates with Microsoft 365 Copilot for spatial productivity and taps Nvidia Omniverse APIs to stream complex 3D digital twins in real time. Why it matters: Apple is positioning spatial computing as a serious enterprise AI tool, moving far beyond casual consumer entertainment.
Google just rolled out Gemini for Google Cloud, infusing its flagship AI across developer, security, and data tools. Duet AI is now Gemini Code Assist, powered by Gemini 1.5 Pro to analyze entire codebases across GitHub and GitLab, while new assistants streamline cloud architecture and BigQuery workflows.
Champions of Otherworldly Magic just revealed it paid an anonymous AI artist $90,000 over six months to produce more than 1,000 card illustrations. The creator worked just 10 hours a month for $15,000, delivering generated and hand-retouched art astronomically faster than traditional teams. Why it matters: Studio-level game art production is rapidly moving from theoretical debates into commercial deployment.
Stability AI just launched Stable LM 2 12B, a multilingual model trained on 2 trillion tokens across seven European languages. Available in both base and instruction-tuned flavors, it is built to handle function calling and RAG workflows on standard hardware. The compact 1.6B model also got an upgrade with better conversation skills and tool use.
Microsoft just announced a new AI hub in central London led by former DeepMind and Inflection scientist Jordan Hoffmann. The team will focus on advancing frontier language models and foundational tooling alongside partners like OpenAI. It directly builds on Microsoft’s £2.5 billion commitment to bolster the UK AI ecosystem.
Spotify just launched AI Playlist in beta, letting Premium subscribers turn random ideas and text prompts into personalized music mixes. You can ask for specific moods, activities, or quirky vibes, and refine the tracks directly in chat. It’s currently testing on iOS and Android for users in the UK and Australia.
Restaurants and grocery stores are now using AI to analyze discarded food and cut down on waste. Startups like Winnow mount cameras over trash bins to track what gets tossed, while Afresh optimizes grocery inventory. Why it matters: The system has helped commercial kitchens slash food purchasing costs by up to 8% by shrinking portion sizes or dropping unpopular items.
Prime Minister Justin Trudeau just announced a $2.4 billion funding package to cement Canada's position in global AI. The bulk of the funds—$2 billion—will go toward expanding computing infrastructure and launching a sovereign compute strategy, alongside $50 million to create a new Canadian AI Safety Institute. Why it matters: As the race for AI compute tightens worldwide, Canada is doubling down on public hardware investment to keep local talent and startups from heading abroad.
OpenAI reportedly transcribed over a million hours of YouTube videos using Whisper to train GPT-4 after burning through high-quality text data. The New York Times reports leadership knew it was legally risky but bet on fair use, while Google and Meta are similarly scrambling in legal gray zones.
Elon Musk just posted on X that Tesla will officially unveil its long-promised robotaxi on August 8th. The autonomous vehicle is reportedly built on Tesla's next-generation platform, as reports surface that the company is shelving its $25,000 mass-market EV to go all-in on self-driving.
Apple is reportedly exploring personal home robotics after ditching its electric vehicle project in February. Engineers are looking into mobile bots that follow users around the house as well as robotic tabletop smart displays. It could give Apple a new hardware vehicle for AI, though any actual product remains years away.
Cisco, Google, Microsoft, and IBM just launched the AI-Enabled ICT Workforce Consortium to tackle AI-driven job disruption. Led by Cisco, the group plans to evaluate AI's impact on 56 tech roles and suggest actionable training programs. Why it matters: With tech layoffs continuing and companies openly planning to replace workers with automation, the industry is trying to prove it can reskill the workforce it's disrupting.
OpenAI just expanded its fine-tuning API and custom models program, giving developers significantly more control over model adaptation. The update introduces epoch-based checkpoints, a comparative Playground UI for head-to-head testing, and assisted fine-tuning with OpenAI researchers. Why it matters: Developers and enterprises can now tailor frontier models to specialized domains with less guesswork and lower compute waste.
Storyblocks just introduced AI-driven search enhancements, leveraging BERT-based models and OpenAI's LLM to deliver fresher and more accurate stock assets. The platform recalculates new content performance every four hours and picks the most engaging video thumbnails, boosting download rates by up to 60%.
Uber and Waymo just launched autonomous food delivery across the Phoenix metro area using Waymo's self-driving cars. Uber Eats customers within the 225-square-mile zone may be matched with an autonomous vehicle, with instructions to unlock the trunk using their phone—though they can still opt for a human courier at checkout.
