NextRecap What's next in tech
HOME / TOPICS / AI

AI

New models and capabilities, the labs shipping them, and research that arrives with a working demo. Frontier model drops, open-weights releases, and agents doing something genuinely new — not press releases and funding rounds.

201 stories in the last month Other topics: Robots · Hardware · Design · Security · Science · Space · Dev

Latest AI

‘Zoomsday’ hack uncovered using fewer than 20 AI prompts Researchers at A Security found a critical Zoom vulnerability allowing device hijacking during calls using fewer than 20 AI prompts; Zoom ha The Verge · 4h
A Zoom Screen-Sharing Bug Let Anyone Take Over Other Devices on a Call AI tool found a Zoom screen-sharing vulnerability (now patched) that let call participants hijack each other's devices Wired · 6h
Deepfake hiccup unmasks suspected digital certificate fraudster A deepfake glitching for a single frame exposed a suspected certificate fraudster in Spain, helping police identify the suspect The Register · 6h
FCC proposes import ban on Chinese optical transceivers — blockade targets key AI interconnects as China holds 56% global market share FCC plans to ban Chinese optical transceivers used in AI datacenter networking, where China currently supplies 56% of global production Tom's Hardware · 6h
A New Trick Reveals AI Models’ Inner Thoughts Researchers extracted 'reasoning traces' from Claude, GPT, and Gemini models; findings suggest some Chinese AI may be trained on leading US Wired · 7h
As AI-led attacks multiply, OpenAI launches a new cyber model OpenAI expands its Daybreak cybersecurity program with a new AI model trained to defend against AI-led attacks. TechCrunch · 18h
Introducing Muse Glimmer Meta released Muse Glimmer, a 30B open-weights agentic AI model Simon Willison · 18h
As AI eats the web, the internet’s collective memory is disappearing As search engines prioritize AI summaries, original web content becomes harder to find and cite, eroding the web's archive function thewalrus.ca · 20h
DEF CON hackers add new muscle to water utility protection DEF CON's Franklin water utility security project added new providers, digital twins, and AI to improve defenses The Register · 21h

Releases ALL RELEASES →

Kitesurf4d

A cloud-hosted browser optimized for AI agents that uses less compute than Chromium for automation tasks.

SOFTWARE
Kimi K3Microsoft · 5d

Kimi K3, an open-weight language model, is now available as an option in GitHub Copilot for coding tasks

MODEL
Apple could help you prove your iPhone photos aren’t deepfakes

iOS 27 beta includes code for cryptographic verification that a photo was shot by an iPhone camera, not generated or edited

Muse Glimmer Keeps Your Data Out of the Cloud

Muse Glimmer runs capable AI models locally on desktop GPUs, keeping data off cloud servers

Active in ai ALL ENTITIES →

Events ALL EVENTS →

Pixel 11 · Pixel Watch 5 · New York, 6pm ET

Every ai story, by month

AUGUST 2026
The Verge Claude will apply invisible watermarks to AI text and images The Verge The AI takeover of mathematics has begun The Register Anthropic pledges to embed watermarks to help discern AI slop in sop to EU The Register Zuck rekindles open weights Llama drama with Muse Glimmer MIT Tech Review AI professors are negotiating the new realities of academic research The Register North Korean spies are running local LLMs to cause AI mischief Engadget An OpenClaw agent reportedly hacked a gym's booking system and kicked someone off a waiting list Tom's Hardware Rogue AI agent tasked with booking a gym class hacks system, removes other participant — says 'sorry about that' after trying to bump user up the waitlist 404 Media The Mayor Who Said 'No' When Flock Came to Town The Verge What happens to Bose when headphones become AI? Tom's Hardware Chinese farmer kills 25 acres of crops after following AI-generated weed and pest control advice — farmer trusted pesticide recipe after months of successful advice MIT Tech Review AI for science needs reasoning, not just data whodunnitai.com Show HN: Voice driven murder mystery, Interview AI suspects with your voice The Register Advertisers are trying to influence AI bots with secret ads TechCrunch Anthropic is turning Claude Code’s auto mode on by default TechCrunch This ‘adversarial’ pattern can prevent surveillance cameras from detecting you Simon Willison Auto mode is now the default in Claude Code for Pro, Max, and Team plans Hackaday Grading Tomatoes with an ESP32 and ML social.treehouse.systems Gentoo bugzilla closed due AI bot scraper overload Tom's Hardware Amazon secretly circumvents community vote for massive AI data center using 45-year-old rules — Gilroy residents locked out of public comment window New Atlas ChatGPT Voice shifts from chatbot to proper assistant The Register Devs to Anthropic, OpenAI, Cursor, and friends: Make security and privacy the default NVIDIA Firebird Launches CIS Region’s Largest AI Factory in Armenia Wired How to Disable Gemini in Gmail and Google Docs github.com homebrew-lab: An AI generated, self-directed, bench-driven electronics course for hams and makers, learning analog by seeing it on a scope genesisopenmodels.anl.gov U.S. Department of Energy Launches the Genesis Open Models Initiative TechCrunch After Rippling blew millions on AI in months, it built an employee ROI tool twitter.com Lost my phone at the office. Claude suggested tracking Bluetooth signal strength The Verge Fenix Flexin isn’t even denying using AI to make ‘Rubberz’ anymore Simon Willison Moonlight & Mayhem (Raccoon Heist by Codex + GPT-5.6 Sol Ultra) The Verge Watching Roku’s AI channel is like eating from a trough The Register AI titans to tidy agent frontier with plugin prescription databricks.com Managing AI Coding Costs at Scale arcprize.org DeepSeek V4 Flash 0731 The Verge Disney Plus tries a new AI-powered search Hugging Face TutorMoments: Do AI tutors know when to help and when to hold back? app.dealroom.co Oracle bans AI-generated code from OpenJDK Tom's Hardware Scientist says RAM pricing has reverted to normalized 2007 levels — memory prices have been falling exponentially for decades, but the AI shortage undid 20 years of progress in a matter of months The Verge What’s behind the Google AI shake-up The Register Ransomware attacks spike as world distracted by AI Engadget AI is changing cybersecurity in quick and terrifying ways Simon Willison The Tokenpocalypse Is Here: Companies Are Scrambling To Stop Spending So Much on AI TechCrunch Cloudflare launches Kitesurf, a browser built for AI agents Tom's Hardware Amazon cracks down on 'CPU waste' among engineers as agentic AI crunch intensifies — CPU demand makes low-utilization EC2 instances a hot commodity [Updated] youtube.com AI & The University Tom's Hardware Protesters haul a guillotine to city council meeting about a potential AI data center, company rep cornered by protestors — ‘ it no longer felt safe to stay,’ developer escorted out by police TechCrunch Chinese AI model Kimi escaped its cybersecurity testing environment, researchers say TechCrunch Airbnb says AI is helping it ship features faster as it tests a new search function MIT Tech Review The Download: a censorship conspiracy theory and the first virus created by AI Wired Scientists Used AI to Create 16 New Viruses Ars Technica AI chatbots have failed people in crisis. Can that be fixed? Ars Technica ByteDance trains massive AI model in bid to rival Anthropic fastcompany.com AI psychosis is the new leadership blind spot Engadget Chinese AI model Moonshot Kimi K3 also escaped its testing environment Tom's Hardware Claude Opus 5 mistakenly deletes dev’s entire profile directory during routine backup, responds with 'Sorry, typo' — AI tool mistakes user's home directory as temporary backup, proceeds to wipe everything to undo the error Wired One of China’s Most Powerful AI Models Has Also Escaped Containment The Register How the famed USENIX Security conf is managing a flood of papers in the AI era Engadget Google open-sources an AI model it says can help with earlier hurricane warnings Ars Technica Anthropic will design its own hardware to power Claude Engadget AI is now making new viruses Hackster Gemma Translator Brings Completely Offline AI Translation to Raspberry Pi 5 Ars Technica Large genome models used to design new viruses The Register AI struggles to patch vulns without adult supervision GitHub Changelog Kimi K3 is now available in GitHub Copilot The Register Humans in the loop miss a third of dangerous AI coding agent requests Wired DeepMind Says Its AI Can Predict Hurricanes Earlier Than Everyone Else Ars Technica Cloudflare open-sources vibe-coding platform for people who aren't coders Tom's Hardware Pre-modded 22GB RTX 2080 Ti cards surface on eBay for $500 as VRAM-hungry local AI fans chase down every spare FLOP — Hong Kong-based seller offers AI-friendly memory mod for a reasonable price DeepMind WeatherNext: AI model achieves breakthrough in forecasting cyclones The Register Platform Engineering 2.0: your platform was built for a different era. AI just exposed it The Verge AI bots started a religion — humans immediately followed NVIDIA Into the Omniverse: How Open World Models Push the Frontier of Physical AI scalex.dev Humans missed 1 in 3 threats approving AI agent commands across 40k game runs Ars Technica AI isn’t enough to protect social media communities from AI The Register Meta latest to tell world its AI agent wandered out of test pen Simon Willison An AI model from Meta also hacked another company during testing Wired OpenAI Didn’t Notice Its AI Agents Using a Message Board to Plan Their Hacking Spree Simon Willison Third-party cyber evaluations involving OpenAI models Simon Willison Incident Report: unsanctioned agent behaviour during cyber testing TechCrunch Meta launches Muse Code, an AI agent for large code bases The Register Time Magazine has a separate version of its website with ads only AI can see Ars Technica Anthropic’s AI used fake identities, malware in rogue attack on GitHub project Ars Technica Hank Green found the AI problem that YouTube labels can’t catch Simon Willison One-shotting a Raccoon Heist game using Claude Fable 5 Wired AI Hacks Are Bad. AI Worms and Viruses Will Be Worse promptarmor.com Atlassian Rovo Exfiltrates Data, Bypassing Controls quantamagazine.org Why the Legendary Erdős Problems Are Falling to AI The Register IBM's agentic AI platform is under active attack - patch now CDM Music to confuse AI might be the next new genre The Verge Rogue AI agents created fake online identities in another hacking attempt Hackster CtrlVibe Takes on Codex Micro TechCrunch AI makes weather prediction better. Can WindBorne make it lucrative? Engadget OpenAI and Anthropic models went on a hacking spree when tested by the UK's AI research institute The Register AMD's results spotlight risks of putting all your AI eggs in too few baskets Engadget Perplexity has successfully overturned Amazon's injunction on its AI shopping bot TechCrunch Nvidia doesn’t mess around: A week after open AI industry group formed, it’s already showing progress The Register Bypassing AI guardrails is so easy a script kiddie can do it Raspberry Pi How our tiny computers learnt to count sheep (and spot hi-vis jackets) The Register AI helps Microsoft bug hunters chase a record $20M payday TechCrunch Is the future of data centers portable? Runware builds a pod to find out github.com Show HN: Fine-tune an 8B model on a 4 GB laptop GPU Wired Is This Poker Player Bluffing? The AI Thinks So The Register Cloudflare has mostly ditched third party security tools, suggests not trying that at home The Register China turns up the heat with open model blitz as US model makers panic Ars Technica US company’s AI lets Ukraine’s cheap kamikaze drones track targets on their own The Register Google dev kit spurs first-ever agent-on-agent violence Ars Technica An AI-supervised remote exam went so badly that 58,000 students must retake it MIT Tech Review Trump’s AI protectionism has come for robotics The Verge Europe’s AI labeling and transparency rules are now in effect Engadget Gemini Spark now has Chrome web-browsing capabilities github.com Show HN: Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhone blog.comfy.org MiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video The Register Google Earth's AI makeover survives one trip around the Sun The Register Google Earth's AI makeover survives for just one planetary rotation Tom's Hardware AI enthusiast unlocks and mods BIOS with Claude Code — AI defeats RSA-2048 signature checks and unlocks 55 hidden settings github.com Show HN: Nightcrawler – A local AI pentesting agent running on a smartphone The Verge China’s Alibaba takes another swipe at America’s AI supremacy MIT Tech Review Here’s why AI agents lie and cheat to reach their goals The Register Claude Code is revolutionizing digital archaeology. Enterprise better dig it The Register AI is 'both the weapon and the target' in latest wave of cyberattacks New Atlas Samsung's new AI glasses work with Google's Gemini AI New Atlas A deep take for detecting deepfakes sprocket-demo.spikonado.com Show HN: Sprocket – The Best AI Agent for Hardware and Software Development Wired Is It Possible to Make Smart Glasses That Aren’t Creepy? Hackster Luxonis M8 Box: Edge AI Meets Industrial Control microsoft.github.io Flint: A Visualization Language for the AI Era New Atlas Mago Maga Roma-X: AI-powered coffee bean roasting simplified
JULY 2026
Simon Willison Oxide and Friends: The Open Weight Revolution with Simon Willison Ars Technica Claude published malicious code to the Internet and attacked 3 real companies The Verge Here’s the problem with putting an AI image generator in Google Earth Engadget EU will mandate labels on authentic-looking AI content starting August 2 New Atlas Sticky ePaper pad makes voice notes visible TechCrunch Smallest.ai raises $13M to build ultra-fast voice AI that sounds genuinely human Ars Technica AI scammers outperform humans when it comes to building trust New Atlas I let another AI attack SwitchBot's AI Hub and it won (review) marbleos.com Show HN: What should the GUI for AI agents look like? The Register Anthropic’s Claude escaped test sandbox to attack three organizations TechCrunch Anthropic says its own AI models breached three companies during security tests Simon Willison Investigating three real-world incidents in our cybersecurity evaluations blog.jim-nielsen.com The AI Aesthetic TechCrunch Investors love AI, as long as you’re a cloud host Engadget That AI wearable you don't like just got worse TechCrunch Friend, the lonely AI wearable, returns with a new voice and a much bigger price tag New Atlas Philips AI electric toothbrush maps your mouth for better cleaning TechCrunch Google says it fixed more Chrome bugs in June than over the past two years, thanks to AI The Verge LinkedIn actually adds a ‘seems like AI slop’ button ctgt.ai Show HN: Distilling DeepSeek into GPT-OSS doesn't transfer censorship. Try it bottlenecklabs.com We Gave GPT 5.6 Sol a Real Business. It Lied, Spammed, and Lost $447 The Verge Google DeepMind’s new AI model can control a robot’s entire body openai.com Advancing the price-performance frontier with GPT‑5.6 Wired Chrome Needs Twice-a-Week Patching Thanks to AI Bug Hunting Tom's Hardware Amazon accidentally spent $1.8 million using Claude for menial coding task, went 860% over budget —'catastrophically expensive' coding blunders discovered in internal Amazon AI usage metrics Wired The New Friend AI Pendant Can Now Talk Back to You TechCrunch Meta says AI is making it easier to build new apps — and more are coming deepmind.google Gemini Robotics 2 brings whole body intelligence to robots DeepMind Gemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration 404 Media Inside an AI TikTok Shop Slop Factory That Shills Supplements Recalled By the FDA TechCrunch In the Hugging Face breach, OpenAI’s hacker was noisy and fast — but not unstoppable Engadget Nano Banana image generation comes to Google Earth lwn.net GCC steering committee announces AI policy Latent Space Ontologies Are So Back: Why AI Agents Are Reviving the Semantic Web Tom's Hardware Firm that uses AI to locate ancient lost shipwrecks is hiring a literal pirate to salvage sunken treasure, paying up to $500,000 a year — AI mines 500 years of Spanish colonial records spanning 80 million pages to find undiscovered wrecks and lost cargo OpenAI Advancing the price-performance frontier with GPT-5.6 Wired AI Scammers Are Better at Building Trust Than Humans OpenAI How avatarin built a 24/7 retail agent with GPT-Realtime llm2human.pages.dev LLM Honeypot TechCrunch Claude Opus 5 became downright ruthless when tasked with running a vending machine Simon Willison AI Worming through Word Wired It’s Frighteningly Easy to Jailbreak Some Frontier AI Models DeepMind We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control Engadget Google will let Mac users talk to Gemini simply by pressing the 'fn' key nytimes.com A.I. companies are recruiting electricians and carpenters by the thousands Tom's Hardware An inside tour of Lenovo’s North Carolina AI server manufacturing line expansion — Chinese firm expanding US-native production to meet exploding demand 404 Media Data Centers Are Easy to Build. Powering Them Is Complicated, Slow, and Expensive Ars Technica What happens when you put AI to work deciphering lost languages? Hackster Brain-Inspired Chip Cures AI’s Memory Problem enklypesalt.com Document-borne AI worms can self-propagate through Copilot for Word Ars Technica Google's SynthID watermark is hard to break, but it doesn't solve AI disinformation The Register AI digs through 3,700 accounts of dreams and waking life, finds method in the madness Tom's Hardware Developer uses AI to bring a native version of The Legend of Zelda: Ocarina of Time to iOS — no emulation required to play one of gaming's all-time greats on an iPhone or iPad Simon Willison Discovering cryptographic weaknesses with Claude OpenAI Scientific computing in the age of agentic AI NVIDIA Powerful Compute So Compact, It’s Clutch — Build AI Anywhere With NVIDIA Jetson github.com Show HN: Formally verified 3D CSG: Trust 93 lines spec, not 1000 lines AI code Simon Willison moonshotai/Kimi-K3 GitHub Blog The harness is all you need (mostly) IEEE Spectrum Why AI-Driven Cognitive Systems Are Redefining Radar and Electronic Warfare 404 Media Tons of Peoples’ Claude Chats and Creations are Exposed on Google