The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
OpenAI launched Patch the Planet, a plan to find hidden software flaws, as an emergency fix arrived for two PHP bugs that could let hackers take over web servers.
Read more: OpenAI launches a plan to find and fix hidden security flaws onlineAnthropic's Claude Code can read, edit and test the files in a software project. It needs a paid Claude plan, so check current prices before you plan around it.
Read more: Claude Code is an AI helper that works inside your software projectTwo customers reportedly bring in about a fifth of Anthropic's sales, and they can leave anytime. That matters if you ever think about buying into an AI IPO.
Read more: Two customers bring in about a fifth of Anthropic's salesA new tool called Opposable lets an AI tap and type on your Android phone, but early tests show it is slow, costly at times and often fails.
Read more: New app lets an AI tap your phone for you, slowly and not always rightUnited Memorial Medical Center in Batavia now uses AI on heart CT scans to judge blockages, which could mean fewer invasive tests and lower costs for patients.
Read more: Batavia hospital uses AI to check heart blockages from a CT scanLegal AI startup Legora raised $550 million at an $8.8 billion value, a sign that law firms are paying for AI tools that handle routine legal work.
Read more: Legal AI startup Legora is now valued at $8.8 billionOpenAI, PerturbAI and Scripps Research say AI agents did in 32 minutes what usually takes human scientists months. The claim has not been checked by outsiders.
Read more: AI agents sift through millions of mouse brain cells in 32 minutesApple posted a short video of the real oil tank and crew behind its new Apple Music look, and one writer says showing human effort now builds trust.
Read more: Apple shows the human hands behind its new Apple Music lookAnthropic's new Claude Duet app works on your email, calendar and other apps in the background, even after you close it. It is invite-only for now.
Read more: Anthropic launches Claude Duet, an AI helper that works in the backgroundWalmart is building a way for AI assistants to shop and buy for you, with spending limits, while Amazon blocks AI shopping bots from its site.
Read more: Walmart Lets AI Assistants Shop for You While Amazon Blocks ThemNvidia signed new deals to secure a base-layer part for its AI chips, which could affect how fast and how cheaply AI tools grow for paying businesses.
Read more: Nvidia locks in supplies of a key part for its AI chipsAnthropic's draft IPO filing shows an $80.7 billion half-year loss, but about $72 billion is an accounting charge, not cash spent running the business.
Read more: Anthropic's Loss Hits $80.7 Billion, Mostly a Paper ChargeA federal judge ruled the EPA cannot end a citizen lawsuit over gas turbines at Musk's Memphis data center, keeping local residents' right to sue intact.
Read more: Judge says EPA cannot shut down a lawsuit over Musk data center pollutionSome TV owners are using an AI helper to turn off ads and tracking on their sets. It may help on some models, but it comes with a safety step.
Read more: TV owners use an AI helper to switch off ads and tracking on their setsFrench startup Pigment raised $250 million at a $2.3 billion value and says it wants to take on Excel and PowerPoint, which matters if you plan budgets in spreadsheets.
Read more: French startup Pigment raises $250 million to challenge ExcelOpenAI is bringing ads to free ChatGPT in 26 more countries, and more than 100,000 advertisers have already run campaigns there.
Read more: ChatGPT ads are spreading to 26 more countriesNews publishers are split on AI: some are suing the companies that build it, while others are signing deals. The outcome could shape what AI knows about the news.
Read more: News publishers split: some sue AI firms, others sign dealsMusic publishers say Anthropic used song lyrics to train Claude without paying. A judge's ruling could shape what AI companies owe creators, and what AI costs you.
Read more: Music publishers say Anthropic's AI copied song lyrics without payingAnthropic's draft bill would let the government block an AI model only for harm over $1 billion or 50 people, a bar that could miss most real attacks.
Read more: Anthropic's AI Shutdown Plan Would Only Kick In at $1 Billion in HarmAndroid Authority reports Gemini will get one hub for apps and skills, so you could control what the assistant can do in a single spot.
Read more: Gemini is getting one place to add apps and new abilitiesOpenAI is adding Locked Chats to ChatGPT, a way to hide private conversations behind a PIN or Face ID. It is not full encryption, so the limits matter.
Read more: ChatGPT adds new protection for your private messagesDemis Hassabis, a Nobel winner who built DeepMind, now says human-level AI could be two to three years away, down from his earlier guess of three to five.
Read more: DeepMind founder now expects human-level AI within 2 to 3 yearsAnthropic signed an AI safety agreement with Singapore and opened a local office, a sign that AI makers and governments are setting testing rules together.
Read more: Anthropic signs AI safety deal with Singapore and opens office thereA writer with work and personal lines on one Samsung phone cut SIM prompts from about 35 to 3 in five days with two settings Samsung barely points out.
Read more: Two hidden Samsung settings cut my SIM pop-ups from 35 to 3
