The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
AT&T is working with OpenAI to make legal answers easier for employees to get. Small businesses with thin legal help may see the same idea soon.
Read more: AT&T works with OpenAI to make company legal advice easier to getOpenAI and Peter Thiel are among backers of a $45 million round for a brain implant meant to fight cancer, a sign tech money is moving into medicine.
Read more: OpenAI and Peter Thiel back a $45 million brain implant to fight cancerWriter Derek Thompson backs building more, but says AI data centers are not like housing and may crowd out hospitals, offices and homes.
Read more: A pro-AI writer says AI data centers are a different matterWhiteLabel.Digital says AI agents will cut its costs and speed up work it does for other marketing firms, with people still checking quality.
Read more: Marketing Firm Uses AI Helpers to Cut Costs for Other AgenciesOpenEvidence, a free ad-funded AI tool, is now valued at $15 billion and widely used by doctors, while studies on its accuracy point in different directions.
Read more: Doctors are adopting an ad-funded AI tool, and accuracy is in questionBank of America says AI assistants that shop for you could cut into marketplaces like Etsy and eBay, and earnings reports next week and the week after may show early signs.
Read more: AI shopping assistants may squeeze online marketplaces like Etsy and eBayNvidia says its new DSX Pod data center can be running in 14 days, a pitch to Europe that could make the computing behind AI faster to build.
Read more: Nvidia's CEO pitches Europe on data centers that arrive ready to plug inGoogle Maps on CarPlay now includes Gemini, so drivers in the US can ask questions and give commands by voice during navigation.
Read more: Google Maps on CarPlay now lets you talk to Gemini while you driveGoogle Maps on Apple CarPlay now lets US drivers talk to Gemini, Google's AI helper, to find places and get directions without touching the screen.
Read more: Google's Gemini voice helper now works in Maps on Apple CarPlayA new French-language prize openly invites fiction written with AI tools, and entrants must say which tool and prompt they used.
Read more: A new book prize is built for stories written with AIIBM released free speech-to-text models that transcribe and translate audio, including one that ranks second among open models on English accuracy.
Read more: IBM releases free speech-to-text models that transcribe and translateMistral and Nvidia plan to use AI to speed up how cars, planes and chips are designed and tested, and to publish open tests so others can check the results.
Read more: Mistral and Nvidia team up to design products faster with AICloudflare says it switched on default protection against two new PHP flaws for all customers about nine hours before the public heard about them.
Read more: Cloudflare blocks two new PHP security flaws before they were publicAnthropic's own study found its AI agents logged six times the hours of its staff, while junior workers' share of new technical hires fell from 27% to 9%.
Read more: Anthropic's AI Helpers Log Six Times Their Human Bosses' HoursRayNeo is selling two kinds of smart glasses in Canada: one acts as a portable big screen, the other translates speech and takes meeting notes.
Read more: RayNeo brings glasses with a giant virtual screen to CanadaThe F Suite launched Floyd, an AI that links every answer to the CFO conversations and documents behind it, so finance leaders can check before they sign off.
Read more: A new AI tool for finance chiefs shows the source behind every answerAnthropic says its Claude AI flagged 493 security issues in 25 open-source projects, including one that sat unseen for 21 years. Here is why that matters.
Read more: Claude found 493 security flaws in 25 open-source projectsAt SaaStr, two AI assistants set up their own shared mailbox and agreed on a safe word, so a human must approve any action. It shows why approval rules matter.
Read more: Two AI assistants built their own shared mailbox and a safe wordSalesforce built a voice AI for business calls that shows its written answer before speaking it, so a person can check it first. The test scores are the company's own.
Read more: Salesforce unveils Launa-Omni, a voice AI built for business callsSafety advocates want OpenAI and Anthropic to cap the computing power used to build new AI, a debate that could shape the tools businesses rely on.
Read more: AI safety advocates want OpenAI and Anthropic to limit computing powerSecurity firm CrowdStrike found four hacking groups using AI tools like Claude to run attacks, and the damage came from old security gaps nobody had fixed.
Read more: Criminal hackers are tricking AI tools into running attacks for themA survey of 1,700 senior managers finds AI is making staff output harder to judge, and many managers are using AI to write the reviews themselves.
Read more: Over half of managers say AI makes it harder to judge staff workServiceNow and Nvidia launched Arc, an AI agent that works on email and Teams tasks and uses layered security checks. Here is what to ask before trusting one.
Read more: ServiceNow and Nvidia launch Arc, an AI assistant for email and TeamsBeyondTrust says its new AI Agent Security tool is available now and checks what an AI helper may do on a work computer before it acts.
Read more: BeyondTrust launches tool to limit what AI agents can do on work computers
