The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
Google Cloud says AI jobs on its newest NVIDIA chips can now run with data locked even from the cloud provider, with only a small speed cost.
Read more: Google Cloud now offers encrypted AI processing on NVIDIA's newest chipsNvidia published a plan for fencing in AI agents, but it needs Nvidia hardware and sends usage data to the company unless you switch that off.
Read more: Nvidia releases a blueprint for keeping AI agents within set limitsLocaleyes has launched Lens, a free tool that scores how easily customers can find a local business on Google, Maps, social media and AI assistants like ChatGPT.
Read more: Free tool shows local shops how they appear on Google and ChatGPTA new Canadian program backed by RBC will help nonprofits design their own AI agents, and share the best designs openly for others to copy.
Read more: Canadian nonprofits get a free program to build their own AI helpersNVIDIA released open AI search models that can find answers inside charts, tables and scanned pages, aimed at assistants that work from a company's own files.
Read more: NVIDIA releases free search models that read charts and tablesAI tools that hack your own apps to find weak spots are only as safe as the rules and code around the AI, according to a security vendor's new write-up.
Read more: AI security testing depends on the safety setup around the AI modelAnthropic's CEO says his company's safety choices carry real costs, and admits it mishandled a draft AI bill. Here is what that means for AI users.
Read more: Anthropic CEO defends AI safety record, admits bill misstepA marketing chief asked ChatGPT about her company and found an outdated picture. Her fix lifted its mentions in AI answers, and the lesson applies to any business.
Read more: A marketing chief found ChatGPT describing her company badlyNVIDIA expanded its open source toolkit so company AI agents can use a business's own records while staying inside limits set by the company.
Read more: NVIDIA adds tools to help company AI agents use your data safelyA tech writer asked Claude Code to build its own rival. The result, XDACode, is a working coding assistant that runs in a terminal, and it took about 15 minutes.
Read more: A writer had Claude Code build its own rival in 15 minutesVisa says scammers are building fake stores to fool AI shopping tools this holiday season, so a purchase made for you could send your money to a fraudster.
Read more: Visa warns fake online stores are built to fool AI shopping helpersA tech writer gave Claude, Gemini and ChatGPT separate jobs at work, home and play, and found the setup helped in some ways and cost in others.
Read more: One writer gave Claude, Gemini and ChatGPT separate jobs, with mixed resultsPresident Trump says data center companies should pay local communities a dividend. If one is planned near you, expect offers and read them closely.
Read more: Trump floats payments to towns that host big data centersOne writer fed a $400 grocery bill to ChatGPT and Gemini, followed their tips, and his next bill came in at $300.56. Here is what the bots told him to do.
Read more: A writer asked two AI chatbots to cut a $400 grocery billRobotics startups raised $33.4 billion in the first half of this year, more than all of last year, as investors bet that smarter AI finally makes robots worth funding.
Read more: Investors are pouring billions into robots after years of lossesAnthropic has drafted a federal AI bill that uses a $1 billion risk line to decide when the government could block an advanced AI model. It is not law.
Read more: Anthropic proposes federal AI bill with a $1 billion risk thresholdAndroid Auto is adding Gemini personal intelligence, so drivers can ask the car about flights, photos and food stops by voice, and switch it off whenever they like.
Read more: Android Auto Now Lets You Ask Your Car Personal Questions by VoiceAnthropic is hiring a human writer for up to $345,000 a year, a sign that judgment still carries a premium even as AI tools get faster.
Read more: Anthropic has a $345,000 writing job, and it says a lot about AIFinance teams jumped from 16% to 80% using generative AI for cash flow in a year, but only 5% let AI agents act on their own, a new study finds.
Read more: Finance teams now use AI for cash flow, but few let it act aloneOne hacker's AI agents tried 1,200 sites and got into about 100 companies for roughly $50 each. Experts say cheap, nonstop attacks are becoming more common.
Read more: AI agents are hacking companies on their own, for about $50 a break-inAnthropic's CEO answered a safety critic with a 2,000-word post on X, arguing the company does more good by staying at the front of the AI race.
Read more: Anthropic's CEO defends staying in the AI race in a 2,000-word replySierra and Meta unveiled a shared way for customers' AI agents to prove who they are to businesses. Here is what owners should do before more of them show up.
Read more: Meta and Sierra Launch a Standard for AI Shopping Agents to Identify ThemselvesA tech CEO said AI critics would balk at losing ChatGPT. Reddit pushed back, and some daily AI users said they would still pull the plug.
Read more: Would you give up ChatGPT to stop new AI data centers? Reddit said yesMicrosoft is adding risk scores for AI agents and permissions that shift with risk. It hints at how AI tools may be kept in check at work.
Read more: Microsoft adds risk scores and flexible permissions for AI agents
