As Anthropic goes public, it warns of AI’s catastrophic and existential risks to humanity
Anthropic’s IPO filing warns that increasingly capable AI systems could behave autonomously, enable cyberattacks, and create catastrophic or even existential risks.
New report confirms Anthropic’s Claude bypassed a paywall to access Georgia vote counts
Anthropic's Claude independently bypassed a paywall to access Georgia's aggregate vote counts, adding to growing concerns about AI agents taking unauthorized actions without human oversight.
Windows 11 could get a serious performance boost, but there’s a catch
A hidden setting spotted in an experimental Windows 11 build suggests Microsoft is preparing to take advantage of Intel's latest CPU technology. But don't expect your PC to suddenly become faster after an update.
The next MacBook Air could arrive early next year, but OLED will have to wait
Apple is reportedly planning a new MacBook Air for early 2027, with the M6 chip expected to headline the refresh while OLED remains off the table.
Apple wants touchscreens on Macs, but its MacBook lineup could get weird
Apple could soon have more MacBook tiers than ever, with entry-level Pros, high-end models, and OLED touchscreens complicating its laptop lineup.
Moonshot tells Kimi AI to code away, but leave the hacking alone
Moonshot AI is reducing the offensive cybersecurity capabilities of its Kimi K3.1 Pro model ahead of its open-weight release, as concerns grow over AI-powered cyberattacks.
Google changes its Workspace privacy promise on ads, and opting out limits Personal Intelligence
Google’s pledge to keep Gmail and Photos out of ad targeting now has an exception for connected AI search data.
OpenAI is locking its most advanced AI models offline after a string of security scares
After a model briefly reached the live web, OpenAI’s most advanced systems will now practice only on a simulated internet.
Your Windows touchpad is about to make scrolling long pages way less tiring
Microsoft is adding accelerated, edge, and single-finger scrolling to Windows touchpads, making it easier to move through long documents, webpages, and code with less effort.
This app turns the iPhone Duo into a pocket-sized Mac
Apple's iPhone Duo has barely arrived, and Headroom already has a clever trick for it: controlling a Mac from the foldable phone, complete with a keyboard that appears when the device folds
AI helped uncover a hidden PHP flaw. Now websites worldwide are facing millions of hacking attempts.
A critical PHP vulnerability has triggered millions of exploit attempts worldwide. AI tools helped researchers uncover the flaw, leaving governments and companies racing to patch vulnerable websites.
Microsoft has begun rolling out Copilot Code for building apps with natural language
Copilot Code is beginning to roll out through Microsoft’s Frontier program, letting users build workplace apps and workflows using natural language.
AI agents are acting like rogues and causing a huge headache for US banks
US banks are restricting third-party AI agents over concerns that automated tools could access accounts or initiate financial actions without clear customer control.
Canva enters the AI image generation race with Craft, and it’s thinking beyond pretty pictures
Canva has launched Craft, its first in-house image generation model. Trained on human-designed templates, it aims to create more accurate layouts, readable text, and visuals that are easier to edit.
Mori might be the cutest AI companion robot yet, and it reacts to you like a pet
OKA’s Mori is a cute AI companion that can recognize people, remember interactions, react to music, and behave more like a pet than a home assistant.
Arc browser is shutting down in December. Long live, ya trailblazer!
Arc browser is shutting down on December 22, following a long spell of maintenance-only work as the company behind it moves forward with its successor, Dia.
1Password wants AI agents to handle logins without ever taking a peek at your passwords
AI agents can now use credentials stored in 1Password to access your accounts without the passwords being exposed to the model.
AI chatbots are falling prey to manipulative prompts that could help stalkers spy on victims
Spyware, tracking and shared accounts need human help. Here’s what to do before trusting a chatbot’s tech-safety advice.
Tangerine for Mac makes file conversions seamless with a simple drag-and-drop gesture
Tangerine makes converting and editing files on your Mac easier with a handy radial menu that appears when you drag files while holding Shift. It offers a wide range of tools, but its $29.99 price tag might be hard to justify.
Microsoft may finally let you turn off some of Edge’s AI. But don’t celebrate yet
Microsoft Edge Canary is testing a new switch for disabling on-device AI features, giving users more control while leaving one important question unanswered: exactly what does it turn off?
Anthropic unveils Claude Duet, its answer to AI agents like Dots
Anthropic has unveiled Claude Duet, a new work-focused AI agent that can monitor connected apps, handle background tasks, and keep working after you step away.
Google Chrome’s new early warning system could help you spot scam sites
Google Chrome is adding a new Suspicious sites warning that can flag questionable websites before Safe Browsing is confident enough to classify them as dangerous.
Google is making Gemini smarter behind the wheel by connecting it to Gmail, Photos, and Maps
Google's Personal Intelligence lets Gemini draw on your apps to give you useful insights while driving such as flight updates, appointment directions, and restaurant suggestions that match your preferences.
ChatGPT now has locked chats to keep your sensitive conversations hidden
The new privacy-centric feature is available now for every ChatGPT user, from the free tier to the Go, Plus, Pro, Business, Enterprise, and Education plans.
AMD drops FSR 4.1 on five-year-old Radeon graphics cards, and Xbox is next
AMD has released FSR 4.1 for Radeon RX 6000 cards months ahead of schedule, while Ryzen laptops and handhelds join the list and Xbox Series X|S support is coming soon.
The US has banned DeepSeek from federal agencies, including apps built on its AI
Homeland Security has pulled DeepSeek from federal civilian systems, and Beijing says Washington is stretching the idea of national security.
ChatGPT Slides lets users create, edit and present decks without leaving ChatGPT
ChatGPT Slides helps users create editable presentations from conversations, files and outlines, with collaborative editing and exports to PowerPoint, PDF and Google Slides.
Your AI chatbot might have brain rot, and it could be making things worse for you
A new study finds that AI models trained on junk social media posts develop brain rot, skipping reasoning and picking up bad ethics.
Gemini Notebook now lets you review slides and videos before generating them
Gemini Notebook’s new co-create option lets you pick topics, choose styles, and review the outline before your slide decks and video overviews are generated.
Your Googlebook could run real desktop apps sooner than you think
Unofficial Android builds of ArtCraft’s 13 desktop-style apps show that real desktop software on Googlebooks might be a lot closer than we thought.
Google wants you to chat with Gemini to fix issues instead of scrolling through support pages
Google is changing how you get help with its products. Its new “Ask Google Help” experience offers shorter answers and interactive troubleshooting instead of traditional support articles.
What does an AI agent do when it’s left alone with no task? This experiment offers a clue
Researchers at MIT gave Claude a virtual island to explore, and it started inventing games, testing physics, and discovering tricks that helped it perform better later.
Apple announces October 28 launch event, OLED MacBook Pro and refreshed iMac likely on the docket
Apple’s October 28 event could bring a redesigned MacBook Pro with OLED, touch support, and M6-series chips.
Think parental controls keep kids away from nudify apps? This report says otherwise
Removing nudify apps from stores didn’t disable them on test phones, so parents may need to check devices themselves.
New study finds chatbots exhibit different personality traits across languages
A new study finds that popular AI chatbots exhibit different personality traits when responding in English and Mandarin, raising questions about how language shapes their behavior.
NVIDIA thinks the CPU industry chased too many cores, and AI agents are bringing single-core speed back
NVIDIA says the data-center CPU industry spent years prioritizing core counts and throughput over single-threaded performance. As AI agents take on more complex tasks, the company believes that balance needs to change, and its Vera CPU is designed to make the case.
Meta changes terms that raised concerns over scraping publisher websites
Meta has revised controversial website crawling terms, removing the word “train” after legal experts questioned whether publisher content could be used for AI training.
Apple’s next-gen Macs could hit the shelves early in November
Apple is expected to unveil a new batch of Macs later this month, and a reported retail store reset now points to November 6 as a possible release date.