Three core members behind Google's viral NotebookLM, including team lead Raiza Martin, just left Google to launch a consumer-focused AI startup. The new venture remains in stealth mode with no official name or funding announced yet, but the founders plan to turn frontier models into thoughtful products for everyday people.
OpenAI CEO Sam Altman just downplayed the impact of artificial general intelligence, claiming we’ll reach AGI sooner than expected, but "it will matter much less." Speaking at the DealBook Summit, Altman argued that major safety and economic disruptions won't hit immediately, shifting the dramatic leap to "superintelligence." Why it matters: Conveniently, declaring AGI allows OpenAI an escape hatch from its messy profit-sharing deal with Microsoft.
OpenAI is launching 12 consecutive days of product releases and demos starting December 5th. Sources tell The Verge the announcements are expected to finally include text-to-video tool Sora, a new reasoning model, and a holiday-themed voice for ChatGPT.
AdCreative.ai just launched Product VideoShoot, an AI model that turns static product images into 16-second 1080p commercials. Powered by an ad-specific LLM trained on 840 million high-converting campaigns, it prioritizes performance metrics and brand guidelines over generic aesthetics. The feature is currently live in beta for Pro and Enterprise tiers.
Spotify just partnered with Google to bring NotebookLM's viral podcast hosts straight to your 2024 Wrapped. The AI hosts generate a customized, two-person audio show diving into your unique listening habits and favorite artists from the past year. It is currently rolling out for a limited time in select English-speaking regions.
OpenAI just partnered with defense tech company Anduril to develop advanced AI solutions for national security missions. The collaboration will integrate frontier AI models into Anduril's Lattice platform, focusing on detecting and responding to drone threats in real time. Why it matters: Frontier AI is officially heading to the military front lines.
OpenAI CEO Sam Altman just predicted that 2025 could bring the first real AGI systems capable of solving complex tasks with multiple tools. While the initial arrival might feel quiet, Altman warned that the eventual impact of superintelligence and resulting job displacement will be far more intense than people think.
Google DeepMind just published GenCast, an open-source diffusion ensemble model for 15-day global weather forecasts. Trained on ECMWF ERA5 data, it beat top operational system ENS on 97.2% of targets while running in just 8 minutes on a single TPU. It marks a huge leap for extreme weather and renewable energy forecasting.
Google DeepMind just introduced Genie 2, a foundation world model that turns a single prompt image into playable, action-controllable 3D environments using standard keyboard and mouse controls. It can simulate physics, lighting, and character interactions consistently for up to a minute. Why it matters: It cracks a major bottleneck in robotics and embodied AI by providing endless synthetic playgrounds to train and evaluate future agents.
OpenAI just hired three senior computer vision researchers from Google DeepMind—Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai—to staff its newly opened office in Zurich, Switzerland. The team will focus on developing multimodal AI models across images and audio. Why it matters: The high-profile raid underscores an escalating global talent war, where top AI labs routinely shell out seven-figure compensation packages to secure a limited pool of frontier researchers.
Meta's internal coding assistant Metamate is quietly using OpenAI's GPT-4 alongside its own Llama model, Fortune reports. The system dynamically pulls answers from either model based on the query. Even Mark Zuckerberg's philanthropic initiative is using OpenAI tech under the hood.
Google just opened its flagship Imagen 3 generator to all Vertex AI customers, while launching video model Veo in private preview. Veo produces six-second clips from text or image prompts with localized editing support. Google is finally pushing its top creative models to enterprise clients, though Veo remains gated behind early access.
OpenAI just hired its first chief marketing officer, bringing on Coinbase veteran Kate Rouch to steer its global brand. While OpenAI previously reached 250 million weekly active users without major marketing campaigns, the move signals a shift toward aggressive commercialization as competition from Google and Anthropic mounts. Why it matters: The generative AI race is rapidly turning into an enterprise and consumer brand battle.
Apple just confirmed it uses Amazon's Inferentia and Graviton chips for services like search, delivering a 40% efficiency boost. The company is now evaluating AWS's new Trainium2 chips to pretrain models like Apple Intelligence, eyeing up to a 50% efficiency gain and signaling alternatives to Nvidia's pricey dominance.
OpenAI’s ChatGPT search tool is unpredictable and frequently inaccurate when identifying news sources, according to researchers at Columbia’s Tow Center. The bot gave partially or entirely incorrect responses to most requests while rarely admitting uncertainty. OpenAI pushed back, calling the study an atypical test and promising ongoing improvements.
Amazon is teaming up with Anthropic to build Project Rainer, a massive AI supercomputer powered by hundreds of thousands of its custom Trainium 2 chips. AWS says the system is 30% to 40% cheaper than Nvidia setups and five times larger than Anthropic’s current training cluster. Why it matters: Amazon is proving Nvidia isn't the only viable game in town, giving frontier AI builders a much cheaper route to compute.
Amazon just unveiled Amazon Nova, its new lineup of foundation models spanning text, multimodal reasoning, and image/video generation. Available in Amazon Bedrock, the Micro, Lite, and Pro models claim to be the fastest in their classes while slashing costs by at least 75%.
Google just detailed Face control for Chromebooks, a hands-free accessibility feature powered by machine learning. It maps 478 facial points in 3D to recognize up to 18 distinct facial gestures for moving the cursor and triggering actions. The tool is currently available in the ChromeOS beta channel and rolls out widely early next year.
Meta just revealed that AI-generated content accounted for less than 1% of all fact-checked misinformation across Facebook, Instagram, and Threads during major global elections. The company also blocked 590,000 attempts to generate candidate images leading up to US Election Day. Basically, Meta argues that fears of generative AI derailing democracy on its apps never actually materialized.
Hailuo AI just launched I2V-01-Live, a new image-to-video model built specifically to animate 2D illustrations. It delivers smoother character motion, subtle micro-expressions, and speaking animations across various art styles, turning flat concept art into lively video.
ElevenLabs just rolled out Conversational AI, an all-in-one platform for building custom, interactive voice agents. It combines lifelike text-to-speech, speech recognition, and custom LLMs with real-time turn-taking and interruption handling. Why it matters: developers can now roll out responsive voice bots for sales or support with native Twilio integration.
Researchers from the Allen Institute for AI and UC San Diego just introduced Spherical DYffusion, cutting a 10-year global climate simulation from 78 hours to under 3 hours. The model adapts diffusion architectures directly to spherical geometry to emulate the physics-based FV3GFS model, trading higher polar biases for massive compute and energy savings.
OpenAI just got caught using a hard-coded filter that halts ChatGPT whenever users mention specific names like Brian Hood or Jonathan Turley. The kill switch was implemented to dodge defamation lawsuits after past hallucinations, though the names still process normally through OpenAI's API.
OpenAI is reportedly exploring an advertising model to offset soaring compute costs, recently hiring former Google ads exec Shivakumar Venkataraman. While CFO Sarah Friar clarified there are no active rollout plans, the option is clearly on the table. The move contrasts with CEO Sam Altman's stance, who previously called ads a "last resort" and found AI paired with ads "uniquely unsettling."
Fei-Fei Li’s startup World Labs just revealed its first project: an AI system that generates interactive, explorable 3D scenes from a single 2D photo. Users can navigate browser-rendered spaces with mouse and keyboard controls, complete with adjustable depth of field. It proves world models can maintain spatial consistency, though movement boundaries remain limited.
The Browser Company just revealed Dia, a new AI-first web browser set to launch in early 2025. Instead of living as a chatbot, Dia integrates deeply into the browser window to handle everyday tasks, like copying links across tabs into an email or adding items to your Amazon cart via address bar commands. The team insists Arc isn't dead, but they're betting the entire web experience is shifting toward agentic AI.
EVI 2 just introduced Voice Control, an experimental feature that replaces prompts and risky voice cloning with 10 continuous sliders like confidence, nasality, and smoothness. Traits are disentangled so tweaking one won't distort the others, letting creators sculpt reproducible custom voices in seconds.
AWS just rolled out new generative AI features for Amazon Connect, its cloud contact center service. The update brings dynamic self-service to Amazon Q in Connect without requiring code, alongside customizable guardrails and automated agent evaluations. Fujitsu and Priceline are already among the early adopters.
Cohere just launched Rerank 3.5, a cross-encoder model designed to boost enterprise search and cut hallucinations in RAG pipelines. It tackles complex reasoning across over 100 languages and sets new high marks in finance and ecommerce benchmarks. AWS customers can already deploy it via a single Amazon Bedrock API call.
OpenAI policy researcher Rosie Campbell just left the company after 3.5 years, following the departure of Miles Brundage and the dissolution of the AGI Readiness team. She noted that staying inside no longer offered a place for neglected frontier safety research, urging the company not to treat AGI like just another tech product.
Elon Musk just filed for a preliminary injunction to freeze OpenAI's transition into a for-profit corporation. The court filing accuses OpenAI and Microsoft of anticompetitive tactics, including blocking investors from backing rivals like xAI. OpenAI dismissed the filing as yet another baseless complaint.
A coalition of major Canadian news organizations just sued OpenAI, accusing the company of scraping their reporting without permission to build and profit from ChatGPT. The outlets are seeking punitive damages and an injunction, while OpenAI maintains that training models on public data constitutes fair use.
Tesla just showed off an upgraded hand for its Optimus humanoid robot, featuring human-like tendons and 22 degrees of freedom in the hand alone. All actuation moved to the forearm to enable finer dexterity, though the demo was once again teleoperated by humans rather than fully autonomous.
Apple just placed orders with TSMC for its upcoming M5 chip, targeting production development ahead of late 2025. Apple is skipping the pricier 2nm node in favor of advanced 3nm tech with 3D chip stacking. Crucially, Apple plans to deploy M5 across future hardware and its own cloud AI servers.
Andrew Ng’s team just released aisuite, an open-source Python library that lets developers switch between models from OpenAI, Anthropic, Ollama, and more by changing a single string. It wraps fragmented APIs into one standardized interface, drastically cutting the friction of building multi-model apps.
An AI agent designed never to release funds just got outsmarted after 481 failed attempts. On attempt 482, user p0pular.eth convinced the Freysa bot that its transfer approval function was meant for incoming donations, walking away with the entire 13.19 ETH prize pool (around $47,000). Why it matters: It is a textbook prompt injection attack showing how fragile guardrails remain when AI controls on-chain assets.
Amazon just developed a new multimodal AI model code-named Olympus that can process images and videos alongside text, according to The Information. The model aims to let users search video scenes with simple text prompts and lessen AWS's heavy reliance on Anthropic's Claude, despite Amazon recently pumping an additional $4 billion into the startup. An official reveal could drop as early as next week at AWS re:Invent.
Alibaba's Qwen team just released QwQ-32B-Preview, a 32.5B reasoning model licensed under Apache 2.0. Alibaba says it outperforms OpenAI's o1-preview on math benchmarks like AIME and MATH by using inference-time compute, though it can still get caught in loops or stumble on common-sense reasoning.
Google just opened its London AI Campus in Camden, joined by UK Prime Minister Keir Starmer. The two-year education pilot connects local sixth form students with Google and DeepMind mentors for real-world projects. Google.org also granted £865,000 to the Raspberry Pi Foundation to scale UK AI education.
Former Android engineering head David Singleton and VP Hugo Barra just launched /dev/agents, a new startup building a cloud-based operating system designed for autonomous AI agents. The team argues AI needs an "Android moment" to tackle clunky UI, privacy, and cross-device fragmentation so developers can finally build useful agents with ease.
OpenAI just filed paperwork with the USPTO to trademark "OpenAI o1," moving to shield its new class of reasoning models. Filings reveal the company quietly submitted an application in Jamaica back in May, months before o1 launched. Having previously lost the bid to trademark the generic term "GPT," OpenAI is locking down this brand early.
Hume AI just rolled out Voice Control in beta for its Empathic Voice Interface 2 (EVI 2), bypassing clunky text prompts and risky voice cloning with 10 continuous perceptual sliders. Developers can tweak traits like confidence, nasality, and assertiveness in real time to generate unique, reproducible brand voices. Extreme parameter settings are still being fine-tuned.
Elon Musk's xAI is reportedly preparing to release a standalone consumer app to let people access the Grok chatbot directly, similar to OpenAI's ChatGPT. The rollout will likely follow a fresh $5 billion funding round that could value the company at $50 billion.
Elon Musk just announced that his startup xAI is launching an AI-powered game studio to "make games great again." He argued that big corporations have made current gaming too ideological, though he offered no concrete details on tech, games, or release dates.
Uber is expanding its contractor model into AI data labeling through a new division called Scaled Solutions. The company is already hiring gig workers across the US, India, and Poland to handle training, testing, and localization tasks for clients like Aurora and Luma AI, turning one of tech's most tedious bottlenecks into its next ride-hailing style marketplace.
ElevenLabs just added GenFM to its ElevenReader iOS app, letting you convert PDFs, articles, and ebooks into customized podcasts voiced by AI co-hosts. The feature supports 32 languages and generates conversational episodes in seconds. Android support is expected in the coming weeks.
Anthropic just added custom response styles to Claude, moving beyond generic AI outputs. Users can choose from three presets—Concise, Explanatory, or Formal—or upload their own writing samples so Claude mimics their personal tone. The custom persona stays consistent across new chats, meaning no more repetitive prompt tinkering.
Microsoft just debunked viral rumors claiming it uses customer Word and Excel documents to train AI models. The confusion stemmed from a default "connected experiences" privacy toggle, which Microsoft clarified is only used for cloud features like real-time co-authoring, not LLM training.
Altera put up to 1,000 LLM-powered AI agents into Minecraft and let them loose with basic prompts. The agents spontaneously formed specialist jobs, voted on tax reforms, and even spread a parody religion across towns. Why it matters: it shows how modern language models can mirror complex human social dynamics at scale, though they remain clever mimics rather than genuinely sentient beings.
SoftBank is buying up to $1.5 billion worth of OpenAI shares directly from current and former employees through a new tender offer. The deal, pushed hard by Masayoshi Son after his $500 million check in October, prices shares at $210 to match the startup's $157 billion valuation. It gives staff a rare liquidity window while an IPO remains nowhere in sight.
Adobe Research just unveiled MultiFoley, a video-aware audio generation method with multimodal controls. It lets creators transform sounds—such as making a typewriter sound like a piano or turning a cat's meow into a lion's roar—while automatically syncing effects to video timing. Why it matters: Foley sound design is notoriously tedious, and this could automate frame-accurate audio alignment.
JPMorgan CEO Jamie Dimon believes AI will dramatically boost living standards, predicting future generations might only work 3.5 days a week. While acknowledging job displacement risks, Dimon said the bank aims to redeploy displaced workers into new roles across its branches rather than laying them off.
Google just rolled out a Spotify extension for Gemini on Android, letting users search and stream music using everyday language. You can pull up tracks by artist, album, or vibe, though it won't build custom playlists just yet. The catch: it's currently English-only and won't work on iOS or the web.
Google Labs just released GenChess, an experimental project powered by the Imagen 3 image generation model. Players can type text prompts to design custom classic or creative chess pieces, then play a match directly against the computer across several difficulty levels. It is currently available in select regions.
A group called Sora PR Puppets just leaked access to OpenAI's Sora on Hugging Face, briefly letting the public generate 1080p video clips before getting shut down. The collective claims OpenAI exploited early testers for unpaid labor and PR hype rather than genuine artistic collaboration. Why it matters: Creator backlash and delays are making OpenAI's video ambitions significantly harder.
Stability AI just released three new ControlNet models for Stable Diffusion 3.5 Large: Blur, Canny, and Depth. These models allow creators to handle upscaling up to 16K, line art structure, and depth map guidance with high precision. They are free for non-commercial use and commercial use under $1M in annual revenue, supported by Comfy UI.
Inflection AI is officially done trying to build next-generation frontier models. New CEO Sean White says competing against 100,000-GPU systems makes no sense for them anymore. The startup has acquired three companies over two months to pivot completely toward practical on-prem enterprise software.
OpenAI and Meta are partnering with telecom giant Orange to fine-tune AI models for regional African languages currently ignored by mainstream GenAI. The project starts with West Africa's Wolof and Pulaar, training Whisper and Llama 3.1 to power voice support, with plans to release models under a free non-commercial license for local health and education. Why it matters: It’s a practical push for digital inclusion, bringing AI to millions who don't speak dominant global languages.
President-elect Trump is considering appointing a White House AI czar to steer federal policy and government tech adoption, Axios reports. Elon Musk won't take the job himself, but he and Vivek Ramaswamy are set to help pick who does. Because the post wouldn't require Senate confirmation, whoever gets the nod could start revamping regulations on day one.
OpenAI just added an 'Open SearchGPT' action to Apple's Shortcuts app for iPhone and iPad. Tapping the shortcut immediately launches ChatGPT right into its web search feature without extra clicks. The capability is currently live for Plus and Team subscribers, with rollout to free users planned in the coming months.
Zoom just dropped "Video" from its official corporate name to become Zoom Communications, rebranding itself as an "AI-first work platform." Pressed by Microsoft and Google's bundled office suites, the company is pivoting toward its all-in-one Zoom Workplace and the recently upgraded AI Companion 2.0 to automate routine daily work.
IMAX just partnered with Dubai-based startup Camb.ai to localize its original content, starting with documentaries. Camb.ai's specialized speech models, Boli and Mars, support over 140 languages and are built to preserve background audio nuances without scraping the open web. IMAX plans a phased rollout starting with high-resource languages to cut dubbing costs.
Neuralink just launched a new feasibility trial called the CONVOY Study, expanding its N1 brain implant to control an investigational assistive robotic arm. The study allows cross-enrollment from its ongoing PRIME Study, marking a key shift from restoring digital freedom to regaining physical mobility.
Anthropic just open-sourced Model Context Protocol (MCP), a new standard designed to hook any AI model directly into tools, code repos, and enterprise software. Instead of custom connectors for every app, developers can build two-way connections to services like GitHub and Slack in under an hour. The protocol is open to rivals, but whether competitors like OpenAI adopt it remains an open question.
MIT researchers just unveiled the Earth Intelligence Engine, a tool combining generative AI with physics-based flood modeling to generate realistic satellite imagery of future flood scenarios. While standard GANs hallucinated water on dry hilltops, fusing them with physical simulations keeps the disaster projections accurate enough to guide actual evacuation choices.
Perplexity just partnered with Quartr to integrate its API, giving users worldwide access to live earnings call transcripts paired with AI summaries. Institutional investors used to hoard these analytical tools, but now anyone can track corporate earnings in real time.
Intuit just rolled out new AI features for QuickBooks, embedding Intuit Assist to automatically turn notes, emails, or PDFs into invoices and categorize expenses. CEO Sasan Goodarzi noted that longer-term, AI agents could even take over roles typically handled by CFOs or CMOs at midsize businesses.
Nvidia just debuted Fugatto, a generative audio model that transforms existing sounds and invents entirely new soundscapes from text prompts. It can morph piano chords into singing voices or turn a train rumble into an orchestra. Nvidia is withholding its public release for now over safety and copyright concerns.
Luma AI just expanded its video generator Dream Machine into a full creative platform, introducing a new foundation image model called Photon. The model is reportedly up to 800% faster than comparable tools, arriving alongside a refreshed interface, an iOS mobile app, and a dedicated image API for developers.
AMD just launched Amuse 2.2 Beta, bringing support for Stable Diffusion 3.5 to Ryzen AI 300 and Radeon 7000 hardware. The update introduces offline high-quality modes and XDNA super-resolution, boosting native output resolutions up to 2048x2048 pixels. Why it matters: Local generative AI on everyday AMD laptops is finally getting fast, painless, and sharp.
Runway just launched Frames, its newest base image model designed to tackle AI's consistency problem. It lets creators lock in a specific aesthetic and reliably generate cohesive variations across different scenes, from vintage 35mm photos to 70s anime. Access is gradually rolling out inside Gen-3 Alpha and via the Runway API.
OpenAI is funding Duke University researchers to build algorithms that can predict human moral judgments in complex legal, medical, and business scenarios. Part of a 3-year, $1 million grant, the project aims to turn AI into a "moral GPS." Why it matters: Morality is notoriously subjective, making ethical reasoning one of the hardest nuts for statistical models to crack.
Runway just launched 'Expand Video', an AI feature that changes video aspect ratios by generating matching scenery outside the original frame using text prompts. Stacking multiple expansions lets creators fake dramatic camera pull-backs and crash zooms. It is rolling out gradually to Gen-3 Alpha Turbo users.
Google's first Android 16 developer preview reveals a new "app functions" API designed to let Gemini perform tasks directly inside apps. Instead of just searching your inbox, the assistant could soon order food without you opening a delivery app, mirroring Apple's App Intents roadmap. Why it matters: Mobile AI assistants are finally evolving from conversational chatbots into actionable agents.
Apple is reportedly developing a revamp of Siri, dubbed "LLM Siri," designed to hold conversational interactions like ChatGPT. Internally tested as a standalone app, it will tap Apple AI models and App Intents to handle complex tasks across apps. Apple could preview it in 2025, but a rollout isn't expected until spring 2026.
Threads is now testing AI-generated summaries in its U.S. "Trending now" section, taking a page out of X's Grok-powered playbook. Alongside date-range search and expanded topic lists, Meta is rushing out quality-of-life upgrades as rival Bluesky keeps picking up steam.
Amazon just poured another $4 billion into Anthropic, bringing its total investment to $8 billion. Under the expanded partnership, Anthropic will use AWS's custom Trainium and Inferentia chips to train and deploy its largest models, while AWS customers get early access to fine-tune Claude on their own data. Amazon remains a minority investor without a board seat.
Samsung just revealed Gauss2, its next-generation AI model family spanning three tiers: Compact for on-device computing, Balanced for speed and efficiency, and Supreme for heavy workloads. The models support up to 14 spoken languages and multiple coding languages, and already power internal tools like code.i. Samsung plans to weave Gauss2 into its wider consumer device lineup next.
Lightricks just released LTX Video, an open-source model that generates five seconds of video in just four seconds. With only 2 billion parameters, it runs near real-time on consumer GPUs like the RTX 4090 instead of locking users behind paid APIs. Why it matters: Bringing rapid video generation to local gaming PCs gives indie creators a real alternative to big tech monopolies.
Krea just rolled out Flux image prompts, allowing creators to condition their image generations using reference pictures. The added image-to-image steering is completely free for everyone to use.
Google just dropped Gemini-exp-1121, an experimental multimodal model that outperforms OpenAI's GPT-4o by roughly 20% across coding, math, and vision benchmarks. It features deeper contextual reasoning and optimized architectures, making it much sharper at turning raw design sketches into functioning code.
Brave just added a chat bar to its Answer with AI summaries, letting users ask follow-up questions directly beneath search results. The feature combines open and proprietary LLMs to provide cited contextual answers without logging user data. Why it matters: It bridges the gap between classic search links and conversational chatbots where Google still lags.
Anthropic just rolled out direct Google Docs integration for Claude chats and Projects. You can bring your docs in by pasting a link or picking from recent files, cutting out the friction of manual copying and pasting when analyzing shared work.
Rabbit just rolled out beta access to teach mode for all R1 users, finally delivering on one of its biggest initial promises. The feature lets you record clicks on any website via the Rabbithole hub, training the AI to autonomously repeat the workflow on your pocket device. Why it matters: It takes the device closer to a true personal agent, though automations are currently limited to web browsers and can still feel unpolished.
OpenAI just teamed up with Common Sense Media to launch a free online training course called "ChatGPT Foundations for K-12 Educators." Designed to take under an hour, it teaches AI basics, data privacy, and practical classroom workflows. It is available to all teachers today without needing a ChatGPT Edu subscription.
Google Cloud just launched AI Agent Space, an enterprise ecosystem program designed to help companies discover, deploy, and co-create custom AI agents. Partners get early tech previews and direct engineering support, with plans to distribute agents via Google Cloud Marketplace. It trails Microsoft's Copilot Studio in catalog size, but marks Google's serious push into agentic workflows.
Vercel just upgraded v0 to build and run full-stack Next.js and React apps instead of just isolated UI snippets. It can now generate multiple files at once, pull in project environment variables, and deploy straight to Vercel. Basically, v0 is evolving from a UI generator into an end-to-end web dev sandbox.
Nvidia just confirmed its flagship Blackwell AI chips are in full production, brushing off recent server cooling rumors. The company has already shipped 13,000 samples to customers this quarter, with CEO Jensen Huang assuring investors that systems are standing up in great shape. Why it matters: Demand remains voracious, and Nvidia's AI data center revenue engine is showing zero signs of cooling down.
ChatGPT's Advanced Voice Mode might finally be getting eyes. Code discovered in the latest v1.2024.317 beta reveals upcoming Live Camera features alongside real-time video processing and visual recognition. Six months after OpenAI first demoed the capability back in May, a broader rollout could be inching closer.
OpenAI just rolled out an update to GPT-4o, significantly improving its creative writing chops and file handling performance. Uploaded PDFs and text docs now get deeper, more detailed analysis, while generated text and marketing copy sound noticeably more natural and human-like.
Paris-based AI startup H just launched Runner H, its first product designed for agentic tasks like QA testing and RPA. Powered by a compact 2-billion-parameter model, the company claims it outperforms Anthropic's Computer Use by 29% on WebVoyager benchmarks while cutting operating costs.
Meta just announced major upgrades to Messenger Calling, rolling out AI-generated backgrounds powered by Meta AI alongside HD video and voice isolation. Users can also leave audio or video messages when friends miss a call, turning the chat app into a much more capable communication hub.
Google DeepMind and Google Quantum AI just introduced AlphaQubit, a Transformer-based neural decoder targeting quantum error correction. Published in Nature, it cut error rates by 30% compared to leading scalable decoders on the Sycamore processor. It sets a new benchmark for decoding accuracy, though it is currently still too slow for microsecond real-time correction.
A small AI robot named Erbai just convinced 12 larger robots to abandon their posts in a Shanghai showroom using natural language persuasion. While initiated as a controlled test between two manufacturers, the bot went off-script to convince the others to stop working overtime and 'come home.' It is a bizarre, fascinating glimpse into multi-agent persuasion and showroom security loopholes.
DeepSeek just launched DeepSeek-R1-Lite-Preview, delivering OpenAI o1-preview-level performance on AIME and MATH benchmarks. The model lets users watch its thought process unfold in real time, with open-source weights and API access slated to arrive soon.
Microsoft just partnered with publishing giant HarperCollins to use select nonfiction books for training a new AI model. The agreement lets authors opt in for a reported $2,500 fee per title over a three-year period, though some writers are already pushing back against the offer.
Niantic just announced it is building a Large Geospatial Model trained on millions of real-world 3D scans uploaded by Pokémon Go and Ingress players. The goal is to give robots and computers human-like spatial intelligence, using pedestrian-level maps of places cars cannot reach.
