The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
Top House Democrats told a court they side with publishers in the New York Times lawsuit against OpenAI, a case that could decide if AI firms must pay for content.
Read more: House Democrats back publishers in copyright fight with OpenAIMastercard, Flybits and Rogers Bank completed Canada's first AI agent purchase, and banks are building checks to make sure the agent really speaks for you.
Read more: AI shopping agents are starting to buy things for you, and trust is the hurdleAn AI agent chose a gene therapy and a brain target, then operated on an epileptic mouse. When it refused to continue mid-surgery, a human had to step in.
Read more: AI picks a gene therapy and performs brain surgery on a mouseAnthropic renamed its two top Claude plans and lifted a usage limit for verified Ultra users, so paying customers should check which plan they are on.
Read more: Anthropic renames its Claude Max plans and drops a usage capA Veracode report finds AI-written code passes more security tests than before, but even the best models still fail more than one in four.
Read more: AI-written software code is getting safer, but still fails oftenDylan Anderson left a job in Minnesota to become the first engineer at Adam, a Y Combinator startup. His advice for anyone eyeing a startup: judge the idea first.
Read more: A Minnesota engineer moved to San Francisco to be a startup's first hireA Seeking Alpha piece says shoppers are starting to chat with AI to find products, but the money from those visits is not yet flowing to everyone.
Read more: AI shopping helpers bring visitors, but few stores are paid yetAmazon's CEO says AI shopping agents don't hurt its ad business, yet Amazon blocked Muse. Here is what that means if you use an AI helper to shop.
Read more: Amazon says AI agents don't hurt its ads, yet it blocked MuseAn unreleased Anthropic AI model found a severe Linux flaw and built a working attack. Some fixes are still arriving, so businesses running Linux servers should check.
Read more: An AI found a serious Linux flaw and built a working attackOpenAI says it will honor a shared robots.txt signal, giving website owners a simple way to say whether their content may be used to train AI.
Read more: OpenAI will respect website owners' choice on AI trainingDoorDash is buying Instacart, and Anthropic is set to go public, per a deal roundup. Delivery app users and sellers should watch for changes.
Read more: DoorDash buys Instacart as Anthropic prepares to go publicAn AI writer says his bet that free-to-download AI models would not cripple the internet looks shaky, as a wave of serious software flaws lands.
Read more: AI writer revisits his bet that open AI models won't wreck the internetAmazon says pharmacy customers grew more than 60% in a year, and its app can now take you from a health question to a prescription at your door.
Read more: Amazon Now Lets You Ask a Health Question and Get the PrescriptionOpenAI rewrote the public rules for how its GPT models should behave, including a ban on resisting shutdown. Whether the models obey is still open.
Read more: OpenAI rewrites its rulebook for how ChatGPT-style AI should behaveZest now lets you text an AI concierge to choose a restaurant and book a table. It works in group chats too, but it learns your taste from your card purchases.
Read more: Text this AI to pick a restaurant and book your tableStarbucks says it will explore drive-thru menus that recognize returning customers and change for them. If you join in, your visits could be tied to your habits.
Read more: Starbucks plans a drive-thru menu that recognizes you and changes for youOpenAI dropped its plan to lower ChatGPT usage limits after users pushed back, so your daily allowance should stay as it is for now.
Read more: OpenAI backs down on plan to cut ChatGPT usage limitsA ransomware group used AI coding agents to take over company networks, which means break-ins may now be faster and need less skill.
Read more: Ransomware gang uses AI coding tools to break into company networksReddit says Google sent it over a million fewer visits a week in October, and its CEO blames Google's AI answers. Any business that relies on search should pay attention.
Read more: Reddit says Google's AI answers are cutting off its visitorsOpenAI says ChatGPT will watch more closely for signs of violence, step in sooner, and publish public safety reports, which matters to any family or team using it.
Read more: OpenAI adds new safeguards to ChatGPT to help prevent violenceThe Justice Department says criminals used stolen identities and AI to pose as college students and collect federal aid. Victims may not know for years.
Read more: Fraudsters use AI to pose as students and steal federal aidTyler Shultz, who exposed Theranos, says AI is hard to check from the outside, which makes it easy for a bad actor to fool investors and buyers.
Read more: Theranos whistleblower says the AI boom is easy cover for fraudAnthropic's Smart Reports lets Claude Enterprise admins see summaries of team chats. It ships off, but if your employer turns it on, your work sessions may be visible.
Read more: Claude's new tool lets your boss read summaries of your AI chatsOpenAI says ChatGPT ads now run in 89 countries, with a wider ad card on desktop and multi-image ads open to every advertiser. Online sellers should take note.
Read more: ChatGPT ads now run in 89 countries and get a wider look
