The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
SK Hynix reported a 632% profit jump and says its memory chips are nearly sold out through 2029, which lifted chip stocks and matters for device costs.
Read more: SK Hynix profit jumps 632% as demand for AI memory chips soarsHome Depot says AI agents can shop and buy on your behalf, but only if they prove they are not just bots and that a real person is behind them.
Read more: Home Depot will let AI shopping agents buy, but they must show IDAmazon now lists 250 million products from rival stores, as AI shopping assistants start to compete for where people begin their shopping.
Read more: Amazon now lists 250 million products from rival storesPrada has launched a ChatGPT tool in the UK, US and Italy that lets shoppers explore products and find stores by chatting, a possible shift away from search bars.
Read more: Prada lets shoppers browse its luxury products by chatting with ChatGPTBlack Duck, a software security company, unveiled an AI agent meant to fix security flaws as quickly as attackers can exploit them. Slow patches put businesses at risk.
Read more: Black Duck launches an AI tool that fixes software security holesA How-To Geek writer says Claude can read a free app's public code and flag danger before you install it. Here is how the method works and its limits.
Read more: You can ask an AI to check free software for danger before you install itAgents working for xAI are reportedly offering to buy homes near a Mississippi power plant, but only if owners agree not to join a lawsuit over noise.
Read more: xAI offers to buy Mississippi homes if owners skip noise lawsuitDiscord will let some public server messages appear in Google from mid-December 2026. Here is who is covered and which setting to check.
Read more: Discord chats from public servers may soon show up in GoogleAnthropic and OpenAI are expected to follow SpaceX onto the stock market, but doubts are growing over whether top AI labs can earn enough to justify the hype.
Read more: Anthropic and OpenAI may go public, but who profits is unclearZoho's Ask Zia now uses a team of six AI agents to analyze business data, and prices stay the same. Here is what small businesses should check first.
Read more: Zoho's business data assistant now uses six AI helpers to do the workFake Claude install guides sit on the real claude.ai site and are promoted through Google ads. Following them can put spying software on a Windows PC.
Read more: Fake Claude install guides are hosted on Claude's real websiteQuake AI, a cloud company with 22,000 AI chips, is building its services for AI agents rather than people. The idea could change how apps and websites work.
Read more: A cloud company bets its future customers will be AI programs, not peopleProphecy has announced DataGPT, a product built to work with Claude, ChatGPT and Microsoft Copilot, so teams may soon use business data inside the AI chat tools they know.
Read more: Prophecy launches tool to use company data inside Claude and ChatGPTProphecy's new DataGPT lets staff run data work inside ChatGPT, Claude and Copilot, with a visible map of the steps behind each answer. A free account is available.
Read more: New tool lets office staff check the steps behind AI spreadsheet answersChatGPT now turns requests for fake election photos into cartoons, which makes them easier to spot before US midterm voters share them.
Read more: ChatGPT now turns fake election photos into cartoonsAnthropic says an unreleased AI model linked software flaws into working attacks on Chrome, Safari, Windows and Linux. All are patched, but updates matter.
Read more: Anthropic's unreleased AI built working hacks for Chrome, Safari and moreDisaster officials may already be using AI chatbots to draft public warnings, and one county's AI-made storm graphic was wrong. Here is what that means for you.
Read more: Emergency officials are quietly using AI chatbots to write disaster alertsA Spanish ALS association and ElevenLabs will give eligible patients a free digital copy of their own voice, so they can keep speaking as themselves.
Read more: Spain's ALS patients can now get a digital copy of their voice for freeAzoma raised $11.2 million for tools that help brands get seen when shoppers ask AI assistants what to buy, which could change how products get found.
Read more: Azoma raises $11.2 million to help brands show up in AI shoppingSymantec captured an attack tool that takes over unprotected AI systems and makes their owners pay for the work. Here is what it does and how to protect your own tools.
Read more: Hackers built an AI agent that hijacks other people's AI systemsFirefox 159 fixes 167 security flaws, 41 of them found by OpenAI. No attacks are reported, but the update is worth checking today.
Read more: Firefox update fixes 167 security holes, 41 found by OpenAIAmazon blocked Meta's Muse shopping agent, and Chase and Citi are limiting outside AI agents. Here is what that means before you give one your logins.
Read more: Banks and Amazon start locking the door on AI shopping helpersAnthropic's new Duet app checks your email and other apps on its own, then sends a morning summary. It asks before it sends, pays, signs or deletes.
Read more: Anthropic's new Claude Duet works on your email without being askedA TLDR AI roundup points to a new Apple model, a Kimi update and a Block voice project, which could shape the apps and payment tools you use.
Read more: Apple's new AI model headlines a busy week of AI news
