Google is cutting ties with Scale AI, its primary data-labeling vendor, after Meta agreed to acquire a 49% stake in the startup. Google had planned to spend $200 million this year, but with rivals like Microsoft and xAI also backing away, top AI labs refuse to risk having proprietary model blueprints exposed to Meta.
Apple just introduced AI-generated App Store tags in the iOS 26 developer beta to boost app discoverability. Instead of basic OCR, Apple uses AI models to pull context from screenshots, descriptions, and other metadata without forcing developers to keyword-stuff their visuals. The tags are human-reviewed, and they haven't rolled out to the public search algorithm yet.
Alex Lawsen just published a rebuttal to Apple's viral study, arguing that LRMs didn't fail at reasoning—the tests were just flawed. Models hit token generation ceilings rather than brain freezes, and even refused mathematically impossible river-crossing puzzles. When asked to write code instead of typing out moves, Claude and o3 solved 15-disk Hanoi puzzles easily.
New York lawmakers just passed the RAISE Act, setting America’s first state-mandated safety reporting rules for cutting-edge AI models costing over $100 million to train. The bill ditches California’s controversial "kill switch" requirement, but tech giants failing to report serious risks still face civil penalties up to $30 million once Governor Kathy Hochul signs it into law.
Tencent Hunyuan just launched Hunyuan 3D 2.1, labeling it the first fully open-source, production-ready PBR 3D generative model. It generates realistic materials like leather and bronze with full physics-based lighting, releases all weights and training code, and runs straight on consumer-grade GPUs.
Google just launched Audio Overviews in Search Labs, using its latest Gemini models to turn search queries into quick, conversational audio recaps. The built-in player also serves up relevant links directly on the results page if you want to dig deeper. It basically brings the NotebookLM audio magic directly into everyday Google searches.
rabbit just dropped its first software-only product, rabbit intern, an AI agent that builds full websites, research reports, and games from a single prompt. It swaps confusing token credits for straightforward task-based pricing, and anyone can test three tasks for free without an r1 device.
Google DeepMind just premiered its first AI-assisted short film, "ANCESTRA," in collaboration with director Darren Aronofsky's Primordial Soup at the Tribeca Festival. The crew paired live-action scenes with the Veo video model, using new motion-matching and object-insertion tools to blend an AI-generated newborn directly into real footage. It marks a big step toward making generative video usable in real Hollywood pipelines.
Nvidia just teamed up with Deutsche Telekom to launch Germany’s largest-ever AI deployment—an industrial AI cloud packing 10,000 Blackwell GPUs in its first phase. Operated by Deutsche Telekom, the setup will power cognitive robotics and simulation-heavy digital twins for European manufacturers, marking a crucial step toward European sovereign AI.
Meta just poured major capital into Scale AI at a valuation topping $29 billion, expanding their data partnership and bringing Scale founder Alexandr Wang over to Meta to lead AI initiatives. Wang will keep a Scale board seat while former Uber Eats leader Jason Droege steps in as interim CEO. Why it matters: Meta locks down high-grade training data pipelines while snapping up one of the industry's most influential builders.
SAG-AFTRA just revealed terms of its tentative video game contract after a 320-day strike over generative AI. Studios must now secure informed consent before cloning a performer's voice or likeness, and using a digital replica for real-time AI chatbots pays at least 7.5 times the minimum rate. Why it matters: It sets the first hard compensation floor on real-time interactive AI clones in gaming.
Betting platform Kalshi just ran a fully AI-generated TV ad during the NBA Finals that cost a mere $2,000 to produce. Created by a solo filmmaker using Google's Veo 3 and Gemini in just two to three days, the project claims a 95% cost reduction compared to traditional commercial production.
A massive internet outage just disrupted major platforms worldwide, including Spotify, Google Cloud, AWS, Cloudflare, YouTube, and Gmail. The downtime hit critical cloud infrastructure and consumer apps simultaneously, knocking key online services offline.
OpenAI just supercharged ChatGPT Projects with support for Deep Research and Voice Mode. The update also improves project memory so it can pull context from past conversations, while mobile users can now upload files and switch models directly in the app.
Microsoft just rolled out Copilot Vision with Highlights for Windows 10 and 11 users in the US for free. Clicking the new glasses icon lets the AI see your selected app, talk through tasks in real time, and highlight exactly where to click next. It is opt-in, doesn't require Copilot Pro, and remains limited to US users for now.
Google just upgraded Gemini Code Assist with its flagship Gemini 2.5 models, boosting reasoning and code generation. Developers can now feed entire folders into chat via a 1M token context window, set project-wide rules, create custom shortcut commands, and switch between multiple saved chat sessions. Personal users can try it for free in VS Code or JetBrains IDEs.
NVIDIA and Stability AI just introduced TensorRT FP8 optimizations for Stable Diffusion 3.5, slashing Large model VRAM needs from over 18GB down to 11GB while boosting performance by up to 2.3x. Developers can also grab the new standalone TensorRT for RTX SDK with JIT compilation. This basically puts heavyweight image generation within reach for standard consumer RTX rigs.
Meta just filed a lawsuit in Hong Kong against Joy Timeline HK Limited, the company behind the CrushAI nudify app, for repeatedly bypassing ad review rules. The company also rolled out new detection tech targeting benign-looking bait ads and shared over 3,800 violating URLs across the tech industry via the Lantern network.
Cisco is rallying behind new president Jeetu Patel, whose aggressive product overhaul and deep Nvidia AI partnerships have pushed shares to 52-week highs. After years of bloated portfolios and sales-first culture, insiders see Patel's technical streamlining as the exact catalyst Cisco needs to modernize for the AI era.
Google DeepMind just launched Weather Lab, featuring an experimental AI model that predicts tropical cyclones up to 15 days ahead across 50 scenarios. Its 5-day track forecasts beat ECMWF's leading physical model by 140 km on average—a 1.5-day leap in warning time. DeepMind is now partnering with the U.S. National Hurricane Center for real-time validation.
Mattel just partnered with OpenAI to bring AI-powered experiences to iconic franchises like Barbie and Hot Wheels. Mattel is also deploying ChatGPT Enterprise internally for creative design, with the duo's first co-developed product set to debut later this year.
Microsoft just added transcript-based video editing to Clipchamp, letting you cut footage simply by deleting words from an AI-generated transcript. It replaces the tedious timeline scrubbing for podcasts and meetings, automatically trimming the video when dialogue is removed. The feature is currently rolling out to work accounts first, with personal accounts coming soon.
Sakana AI just unveiled Text-to-LoRA, a hypernetwork that generates task-specific LLM adapters directly from plain English prompts. Instead of gathering datasets and burning compute on fine-tuning, it creates LoRAs in a single step and even generalizes to unseen tasks. The code is open source ahead of ICML 2025.
Elon Musk just shared that Tesla tentatively plans to kick off its robotaxi service on June 22 in Austin, Texas. The rollout will start small with 10 to 20 Model Ys across a limited radius, though Musk warned the date could shift as the team remains paranoid about safety.
Disney and Universal just sued Midjourney in a major legal showdown, calling the AI image generator a "bottomless pit of plagiarism" for churning out unauthorized copies of Darth Vader, Shrek, and Minions. The entertainment giants also targeted Midjourney's upcoming video generator and demanded a jury trial.
The Wikimedia Foundation just halted a mobile test of AI-generated article summaries following fierce backlash from Wikipedia's editor community. Contributors argued that chasing Google's flashy AI wave insults readers and threatens to inflict irreversible harm on Wikipedia's hard-won reputation as a trustworthy, serious source.
Meta just rolled out generative AI video editing across the Meta AI app, Meta.AI website, and Edits app. Users in the US and over a dozen countries can choose from over 50 preset prompts to transform 10 seconds of video into comic styles or video games for free. Meta says the tool was inspired by its Movie Gen research, with custom text prompts coming later this year.
The Browser Company just launched Dia, a new AI-first browser now in beta for existing Mac Arc users. Dia returns to a familiar Chrome-like layout but embeds an AI assistant directly into the sidebar that can read your open tabs, cookies, and browsing history. Why it matters: it bets that your browser should be your primary everyday AI companion.
Mistral AI just unveiled Mistral Compute, a private full-stack AI infrastructure platform spanning GPUs, orchestration, and APIs from bare metal to fully managed PaaS. Partnering with NVIDIA, it aims to offer an independent, sovereign European alternative to major US and Chinese cloud providers.
DeepL just deployed Nvidia's DGX SuperPOD packed with B200 Grace Blackwell chips, cutting the time needed to translate the entire internet from 194 days to just 18. The $2 billion startup plans to leverage the massive compute boost to train next-gen translation models and power tools like Clarify.
Samsung just rolled out a multi-voice recognition update for its 2025 Bespoke AI fridges. Powered by Bixby, the built-in smart display can now identify different family members by voice to automatically pull up personal calendars, photo albums, or matched phone accessibility settings. It is a clever way for Samsung to deliver personalized smart home experiences without a hit smart speaker.
Starbucks is piloting an AI virtual assistant called Green Dot Assist across 35 stores to support baristas in real time. Powered by Microsoft Azure's OpenAI platform, the in-store iPad tool answers recipe questions, troubleshoots espresso machines, and helps managers cover shifts. A wider rollout across the U.S. and Canada is planned for fiscal 2026.
Meta is reportedly setting up a brand-new AI research lab dedicated to pursuing superintelligence. As part of a broader shakeup, Mark Zuckerberg has tapped Scale AI CEO Alexandr Wang and is in talks to invest billions into his startup, while dangling seven- to nine-figure pay packages to poach researchers from OpenAI and Google.
Sam Altman just confirmed OpenAI is pushing back the launch of its open-weights model past June to later this summer. Altman teased that the research team pulled off something unexpected and amazing, making the delay well worth the wait. It looks like OpenAI's first open-weights release might pack bigger surprises than anticipated.
OpenAI CEO Sam Altman just declared that humanity has passed the "event horizon" toward superintelligence in a new essay. He predicts AI will generate novel scientific insights by 2026, followed by real-world robotics in 2027, eventually driving intelligence costs down to near the price of electricity. Why it matters: Altman is framing an earth-shattering shift as a routine transition, while dropping clear hints about OpenAI's next frontier.
OpenAI just rolled out o3-pro, an upgraded reasoning model that replaces o1-pro for ChatGPT Pro and Team subscribers, as well as API developers. It brings built-in tool access like web browsing and Python execution, beating rivals on tough math and science tests, though it takes longer to respond and skips image generation for now.
Snap just announced plans to sell its lightweight AR glasses, Specs, to everyday consumers starting in 2026. The new see-through frames pack an onboard multimodal AI assistant to process video and audio, alongside developer support for OpenAI and Google DeepMind models.
MIT researchers just monitored brain activity during essay writing and found that ChatGPT users had significantly weaker neural connectivity than those using search engines or no tools at all. When forced to write unaided later, former AI users showed lower brain engagement and struggled to recall their own work. Why it matters: while AI tools save immediate time, heavy reliance might quietly hollow out critical thinking skills.
Krea AI just launched Krea 1, an image generation model built specifically to avoid the dreaded plastic AI look with realistic skin textures and dynamic framing. It outputs native 1.5K images upscaled to 4K, generates HD visuals in seven seconds, and lets creators fine-tune custom styles with up to 50 uploads. Daily free tiers are available without requiring a credit card.
Google's AI Overviews and new chat-driven search tools are devastating referral traffic for digital publishers, according to a Wall Street Journal report. Users now get quick answers directly from chatbots instead of clicking blue links, sending The New York Times' organic search traffic share down to 36.5% in April. To survive, media outlets are rushing into licensing and revenue-sharing deals with AI giants.
Microsoft just rolled out an AI-powered search for browsing history in Edge Beta. It can find previously visited pages even if you use synonyms, vague phrases, or typos. Best of all, it runs entirely on an on-device model and your data never leaves the machine—sidestepping the privacy backlash seen with Recall.
Mistral AI just introduced Magistral, its first-ever reasoning model family designed for transparent, multi-step problem solving. It comes in two flavors: an open-weight 24B model (Magistral Small) and an enterprise version (Magistral Medium). Why it matters: it reasons natively across multiple languages and claims up to 10x faster throughput via Flash Answers in Le Chat.
OpenAI just restored most of its services following a massive daylong outage that crippled ChatGPT, Sora, and its developer APIs. While the API and core chatbot features are back to normal, OpenAI says voice mode is still suffering from elevated error rates.
Cisco just overhauled its networking portfolio across data centers and enterprise campuses to address the crushing bandwidth demands of modern AI workloads. The rollout includes a unified Nexus Dashboard equipped with a troubleshooting AI assistant, new 400G optical modules, and SiliconOne-powered campus switches. Cisco is positioning itself as the complete plumbing provider for the agentic AI era.
Qualcomm just unveiled its Snapdragon AR1+ Gen 1 processor at AWE USA, shrinking chip size by 26% while powering on-glass generative AI without relying on a phone or cloud connection. During a live demo on RayNeo X3 Pro frames, it ran a local Llama 1B model entirely offline, taking smart glasses closer to true standalone devices.
Google just launched Veo 3 Fast inside the Gemini app and its Flow filmmaking tool. The new model doubles generation speeds while keeping the same 720p resolution, alongside server optimizations to handle massive traffic. Why it matters: Faster turnarounds make rapid prompt iteration much less painful for video creators.
Apple just launched the Foundation Models Framework at WWDC, giving third-party developers direct access to the on-device models behind Apple Intelligence. It supports offline tasks and tool calling with zero inference costs, with a public beta rolling out next month.
Anthropic just quietly shut down "Claude Explains," an experimental blog launched about a month ago where Claude drafted technical guides with human oversight. The site was wiped and redirected after social media pushback over vague AI disclosure, looking to many like a generic automated marketing push.
Top Chinese AI apps like ByteDance’s Doubao, Alibaba’s Qwen, and DeepSeek just disabled photo-recognition and exam-solving features during the country's college entrance exams. Chatbots told users the temporary blackout was implemented to prevent cheating. Basically, an AI kill-switch for exam season.
Tesla’s humanoid robot engineering lead Milan Kovac just announced his departure from the company after a decade, with Autopilot chief Ashok Elluswamy reportedly taking over. Why it matters: Elon Musk has pegged Tesla's future on building 5,000 Optimus bots this year, making this high-profile leadership shakeup a major test for its robotics timeline.
Apple researchers just introduced STARFlow, an image generation system combining normalizing flows with autoregressive transformers to rival DALL-E and Midjourney. Unlike iterative diffusion models, it enables exact maximum likelihood training in continuous spaces. Why it matters: it gives Apple a unique technical path that could play to its strengths in on-device AI and fine-grained control.
Stability AI is in hot water as Getty Images takes it to London's High Court in the first major copyright trial of the generative AI era. Getty argues the startup brazenly scraped its photo library to train Stable Diffusion without paying a dime, while Stability claims the case threatens its business and the broader AI sector. Why it matters: This trial could reshape content licensing terms for every image generator on the market.
Apple just introduced visionOS 26 for Vision Pro, adding persistent spatial widgets and a generative AI algorithm that converts regular photos into interactive spatial scenes. The update also delivers sharper Personas with full side profiles, in-room shared experiences, and support for PlayStation VR2 Sense controllers. Developer betas are available now.
Apple just upgraded Apple Intelligence across its lineup, introducing systemwide Live Translation and on-screen visual intelligence. Crucially, third-party developers can now tap directly into Apple's on-device foundation model for free offline inference via Swift. Public rollout arrives this fall.
OpenAI just rolled out an upgrade to ChatGPT's Advanced Voice mode for paid subscribers, making spoken conversations sound remarkably human. The update brings subtler intonation, realistic pauses, and emotional nuances like sarcasm and empathy, plus a hands-free continuous translation feature. Keep in mind that minor pitch glitches and occasional background hallucinations can still pop up.
Anthropic just appointed Richard Fontaine, CEO of the Center for a New American Security, to its independent Long-Term Benefit Trust. Fontaine brings decades of defense and foreign policy experience to help guide the company through AI-driven geopolitical risks. The independent trust oversees board selection without holding any financial stake in the startup.
Microsoft just rolled out AI-powered Relight and natural language search to the Photos app on Windows 11 Copilot+ PCs. Users can place up to three customizable virtual lights on any shot and search their local library using plain descriptions like "sunset at the beach." For now, Relight is rolling out to Insiders on Snapdragon PCs, with AMD and Intel versions coming in the next few months.
EleutherAI just released The Common Pile v0.1, an 8-terabyte AI training dataset built entirely from licensed and public domain text over two years. Alongside it, the team dropped two 7B Comma models that perform on par with early Llama models, pushing back on the industry narrative that frontier AI requires unpermitted scraping.
Google just launched scheduled actions inside the Gemini app, allowing users to automate routine prompts. You can now tell Gemini to run tasks at specific times or set up recurring workflows, like a Monday morning brainstorming session or daily calendar briefings. Why it matters: AI assistants are finally shifting from purely reactive chatbots into proactive task runners, though it currently requires a Google AI Pro, Ultra, or qualifying Workspace plan.
Apple is set to expand Apple Intelligence at WWDC 2025 with major upgrades to translation and developer tooling, according to Bloomberg. Users could get automated Message translation, live phone call translation, and real-time conversation translation right through AirPods, alongside new APIs opening Genmoji-style tools to third-party apps.
Anthropic just introduced Claude Gov, a specialized suite of models built exclusively for U.S. national security agencies in classified environments. The models are fine-tuned to refuse less when handling sensitive intelligence and cyber data while keeping rigorous safety guardrails intact. Why it matters: Frontier AI labs are officially racing into the defense sector with high-clearance, custom-built AI.
ElevenLabs just rolled out the alpha release of Eleven v3, its upgraded text-to-speech model featuring heightened emotional awareness and context adaptability. The model supports over 70 languages, up to 5,000 characters per request, and bracketed audio tags for directing whispers and laughter. Why it matters: creators and developers can now generate nuanced, natural character dialogue without hiring voice actors.
Higgsfield AI just introduced Higgsfield Speak, a new tool built to create motion-driven talking videos quickly. Users pick a style, select an avatar, and type in a script, while the platform automatically generates cinematic motion, voiceover, and facial emotion.
X just updated its developer agreement to bar third parties from using platform posts or its API to train or fine-tune foundation models. The restriction mirrors Reddit’s playbook and could pave the way for lucrative data licensing deals. Meanwhile, X is still freely feeding user data into its own Grok AI.
Perplexity just integrated the SEC/EDGAR database across Search, Research, and Labs for all users. The update gives everyday investors direct access to comprehensive financial filings, turning dense regulatory paperwork and technical disclosures into readable insights on demand.
Google just rolled out an upgraded preview of Gemini 2.5 Pro before its general availability in a couple of weeks. The update gained 24 Elo points on LMArena to hit 1470, improves formatting and creative writing, and introduces thinking budgets in Google AI Studio to let developers manage latency and cost.
HeyGen just rolled out a major update to Avatar IV, adding script-driven dynamic gestures and hyper-real micro-expressions to its digital avatars. Users can now direct body movements directly via prompts in 1080p clips up to 60 seconds long. The feature is already available across all HeyGen accounts.
Stanford researchers just released Tokasaurus, an open-source LLM inference engine built specifically for high-throughput batch workloads like synthetic data generation. By slashing CPU overhead, dynamically grouping shared prefixes, and optimizing pipeline parallelism for non-NVLink GPUs, it beats vLLM and SGLang by over 3x on key benchmarks.
Voice AI startups raised $2.1 billion last year, an eightfold spike according to CB Insights. Fueled by real-time breakthroughs from OpenAI and ElevenLabs, full-duplex agents are cheap enough and realistic enough to take over drive-thrus, clinics, and call centers where callers can barely spot the bot.
HubSpot just rolled out a deep research connector for ChatGPT, letting paid ChatGPT users analyze their own CRM data without leaving the chat. The integration mirrors existing HubSpot access permissions so reps only see their own accounts, and HubSpot confirmed data won't be used to train OpenAI's models.
Cursor just launched Cursor 1.0, giving the AI editor built-in code review, the ability to remember past mistakes, and background multitasking for dozens of jobs at once. It means developers can offload tedious fixes and refactoring to the background without getting blocked.
OpenAI is pushing back against court demands driven by The New York Times requiring it to indefinitely retain consumer ChatGPT and API logs, including deleted chats. OpenAI argued the sweeping order directly breaches its 30-day deletion privacy commitment and confirmed the disputed logs remain locked down from plaintiffs.
OpenAI is fighting a court order requiring it to preserve all ChatGPT logs, including deleted chats and sensitive API data. The mandate followed accusations by The New York Times and other publishers that OpenAI was destroying copyright evidence. OpenAI called the sweeping order premature and warned it puts hundreds of millions of users' privacy at risk.
Reddit just sued Anthropic, alleging the AI lab crawled its site more than 100,000 times since July 2024 after claiming it had stopped. Reddit argues its two decades of authentic human conversations were commercially exploited to train Claude, while Anthropic denied the claims and promised a vigorous defense.
Microsoft just launched Copilot Shopping across its web and mobile apps. It acts as an end-to-end buying companion that curates product cards, compares cross-retailer prices, tracks historical deals with custom alert sliders, and enables native checkout. Why it matters: Conversational agents are moving from answering questions to managing the entire consumer wallet.
OpenAI just updated ChatGPT's deep research mode to pull real-time context directly from internal workspace tools while respecting existing user permissions. The new connectors link with Outlook, Teams, Google Drive, Gmail, and Linear. It is rolling out to Team, Enterprise, and Edu accounts, as well as Plus and Pro users outside the EEA, Switzerland, and the UK.
Anthropic just rolled Claude Code into individual Pro and Max subscriptions, letting developers run the agent directly in terminal and IDEs like VS Code or Cursor without separate API billing. Usage counts toward your shared 5-hour and weekly plan limits, though having an API key set in your environment will still trigger standard token charges.
Luma AI just launched Modify Video, a video-to-video editing tool inside Dream Machine: Ray 2 that reimagines backgrounds, lighting, and characters while locking down original motion, facial expressions, and camera paths. In blind evaluations, Luma claims it consistently beats Runway V2V on temporal consistency and motion retention. It is a huge shortcut for creators who want rapid iterations without reshooting from scratch.
Mistral just launched Mistral Code, an AI coding assistant built on the open-source Continue project to take on Cursor and GitHub Copilot. It bundles in-house models like Codestral and Devstral, with a private beta now rolling out for VS Code and JetBrains development platforms. The big draw for enterprise teams is deployment flexibility: companies can run it in the cloud, on private servers, or entirely air-gapped.
Epic Games just revealed plans to let creators build voice-interactive AI NPCs inside Fortnite, following the massive success of its AI Darth Vader. Creators will get access to the same safety guardrails Epic used to keep Vader from going rogue, with live demos showing dynamic voice conversations. Why it matters: Bringing unscripted, talking agents into custom worlds could turn static game maps into living playgrounds.
Runway just partnered with AMC Networks, the studio behind Breaking Bad and The Walking Dead, to integrate its AI tools into marketing and TV development. AMC plans to use Runway's models for promotional campaigns, VFX ideation, and accelerating pre-visualization during series creation. Why it matters: Entertainment powerhouses are actively embedding generative video tools straight into their core production pipelines.
Samsung just teased the next chapter of its Ultra lineup, promising to pack its top-tier specs and foldable-optimized Galaxy AI into a lighter, more compact body. The device promises voice-driven multitasking on the go and adaptive AI workspace features when unfolded. Ultra is finally going folding with smarter onboard models.
Figma just launched the beta of its Dev Mode MCP server, linking design context directly into AI coding agents like Cursor, Windsurf, and Claude Code. Instead of relying on flat screenshots, LLMs can now pull exact components, tokens, and code syntax straight from Figma files. Why it matters: It brings design systems directly into the agentic coding loop.
Anthropic just pulled the plug on Windsurf's direct access to Claude 3.x models with less than five days of notice. Co-founder Jared Kaplan admitted it would be 'odd' to supply Claude to OpenAI amid rumors of a $3 billion buyout. Windsurf has scrambled to third-party inference providers to keep users coding, but short-term hiccups may follow.
Suno just rolled out a major upgrade to its Song Editor, giving creators granular control right from the waveform. You can now reorder and rewrite tracks section by section, split audio into 12 clean stems, upload songs up to 8 minutes long, and fine-tune output vibes using creative sliders.
Builder.ai just filed for bankruptcy after revelations that its $1.5B software platform was actually powered by hundreds of engineers in India pretending to be bots. Internal documents exposed years of manual coding disguised as proprietary AI, alongside inflated revenue schemes and massive unpaid cloud bills. Why it matters: It marks the industry's biggest "AI washing" collapse since ChatGPT, triggering federal fraud probes and sending shockwaves through tech investors.
Anthropic quietly rolled out Claude Explains, a technical blog mostly written by Claude to showcase its drafting chops. But it's not fully automated: human editors and subject-matter experts review and polish every post before publishing. Anthropic insists this is a demo of human-AI collaboration, noting it's still actively hiring human writers.
OpenAI just published an Outbound Coordinated Disclosure Policy to govern how it reports vulnerabilities discovered by its AI tools in third-party software. With its models already catching zero-days, OpenAI is defaulting to open-ended fix timelines rather than strict deadlines to give developers breathing room.
Google just quietly hit pause on rolling out its AI-powered Ask Photos search feature. A product manager admitted the tool isn't where it needs to be due to latency, quality, and UX problems, promising a revamped version with better speed and recall in about two weeks.
OpenAI just expanded ChatGPT's memory to free users, letting the chatbot reference recent chats alongside saved notes for more tailored answers. Paid tiers get long-term retention while free users get lightweight continuity, and you can wipe specific details or turn off history reference in settings anytime.
OpenAI just started rolling out a lightweight version of memory improvements to free ChatGPT users. In addition to saved memories, the bot can now reference your recent chats to give more personalized responses. It makes casual writing, advice, and learning feel much more continuous without paying for Plus.
Google just rolled out public link sharing for NotebookLM, letting anyone browse and interact with your curated notebooks via a single URL. Viewers can ask questions and listen to generated Audio Overviews without touching source files. For now, it is limited to personal accounts.
ManusAI just launched its new video generation feature, turning single text prompts into fully structured, multi-scene video stories in minutes. The tool automatically plans scenes, crafts visuals, and animates the story. It is currently available in early access for Basic, Plus, and Pro subscribers, with wider access coming soon.
Wispr Flow just launched an iOS app that functions as a system keyboard, letting you voice-type inside any app across more than 100 languages. It learns custom vocabulary, works in poor network conditions, and even picks up quiet whispers. Users get 2,000 words free per week, with an unlimited tier priced at $12 a month.
AI phone support startup Phonely just teamed up with Maitai and chipmaker Groq to slash voice bot response times by over 70%, pushing time-to-first-token down to 176 milliseconds. The system hit 99.2% accuracy across iterations, and over 70% of callers now can't tell it's a machine—prompting one client to swap out 350 human agents this month.
TikTok just introduced AI-powered Smart Keyword Filters to expand content controls on the For You feed. The tool uses AI to automatically detect and block synonyms and related phrases alongside user-submitted words. TikTok also plans to double the keyword limit to 200 and add bulk input options in the coming months.
Amazon MGM is fast-tracking Artificial, a feature film dramatizing the wild 2023 OpenAI boardroom coup that briefly ousted CEO Sam Altman. Luca Guadagnino is in talks to direct, with Andrew Garfield reportedly circling the lead role and production eyeing a summer start.
The Alan Turing Institute just surveyed UK families and found that 22% of children aged 8–12 have used generative AI, with ChatGPT being the clear favorite. Most kids turn to these tools for creative play and homework, while children with additional learning needs rely on them heavily for social connection and advice.
AI pioneer Yoshua Bengio just launched LawZero, a non-profit backed by roughly $30M to build guardrails against deceptive autonomous agents. Its upcoming "Scientist AI" acts like an impartial psychologist, outputting probability scores instead of confident answers and blocking actions likely to cause harm. Why it matters: As frontier models gain agency and self-preservation tendencies, we need watchdog systems at least as smart as the models they police.
