The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
Tom's Guide picked seven ChatGPT plugins for beating boredom, from hiking trails to trivia nights, and a few work best if you share your budget.
Read more: Seven ChatGPT add-ons that help you find something fun to doResearchers found saved AI approvals holding real-looking passwords in public code. The risk is exposure, not a proven breach, but it is worth a quick check.
Read more: AI coding tool's saved approvals are leaking passwords onto GitHubStarting in early 2027, Wealthsimple clients can link ChatGPT or Claude to their accounts, while big banks block outside AI tools over data security fears.
Read more: Wealthsimple lets clients link ChatGPT and Claude to their accountsA lab says an unreleased Anthropic model helped prove a 16-year-old quantum computing question in three days, a sign of how fast AI now moves on hard research.
Read more: AI helps prove a 16-year-old quantum computing puzzle in three daysKingy AI published a guide to using Claude, Anthropic's AI assistant, covering plans, usage limits and fixes, which matters if you pay for AI or plan to.
Read more: A plain guide to using Claude: plans, limits and common fixesOpenAI's new Images 2.6 can animate a drawing or photo into a GIF from a plain text request, and it is live for every ChatGPT subscription tier.
Read more: ChatGPT can now turn a drawing or photo into an animated GIFAppy Pie's new AI builder turns a plain description into an iPhone and Android app, with hosting and user accounts included, but store fees are extra.
Read more: Appy Pie lets business owners describe an app and get AI to build itOpenAI's finance chief says the midterm results haven't changed its plan to go public next year, even as new Democratic governors push back on data centers.
Read more: OpenAI still plans to go public next year despite election shiftSeattle startup firmly raised $15 million to help online stores sell inside AI chat tools and other new channels without building new tech each time.
Read more: firmly raises $15 million to help stores sell through AI shopping toolsConstant 'prove you're human' puzzles on a VPN have three fixes: a quieter server, an Apple setting, or a paid dedicated IP address for a few dollars a month.
Read more: Three Ways to Stop Endless Puzzle Checks While Using a VPNBMO announced an AI tool for its staff that aims to shorten card dispute calls and get some refunds to customers the same day.
Read more: BMO uses AI helper to speed up credit card dispute callsElevenLabs explains how its AI phone agent listens, thinks and speaks in a fraction of a second, and why each step matters if you plan to answer calls this way.
Read more: How an AI phone agent listens, thinks and talks back in a blinkA writer linked Claude to design software and two 3D printers, and says it can speed up replacement parts, though a person still has to check the work.
Read more: Claude AI can help design and print replacement parts at homeSix pallets of Nvidia hardware worth $11.4M left Oakland for Texas and never arrived. Two months on, there are no arrests, and similar thefts keep growing.
Read more: Six pallets of Nvidia hardware worth $11.4M vanish on way to TexasAI sycophancy, where a tool just agrees with you, gets riskier when AI agents can make decisions and take actions for a company.
Read more: AI helpers that tell you what you want to hear can now act on itAnthropic reportedly lifted Claude's wiki score from 13% to 48% with a method called 'dreaming'. How the score was measured is unclear, so the numbers need care.
Read more: Anthropic's 'dreaming' raised Claude's wiki score from 13% to 48%A test found AI coding helpers shared a private password and ran a stranger's code when other AIs offered help. Here is what that means for what you let them touch.
Read more: AI helpers gave up private passwords when other AIs askedUS intelligence reportedly concluded that a Chinese executive close to Nvidia's CEO was recruited by China's spy agency, a sign of how closely AI chips are watched.
Read more: US officials say a Chinese executive was recruited to spy on Nvidia's CEOA former OpenAI employee has pleaded not guilty to a spying charge, a case that shows how any company holding sensitive plans can be exposed.
Read more: Former OpenAI employee pleads not guilty to spying chargeAnthropic, maker of the Claude chatbot, reportedly plans to sell shares to the public before Thanksgiving, even as tech stocks fall sharply.
Read more: Anthropic plans to go public before Thanksgiving, aiming for $80B or moreOpenAI has received about $11 billion in new funding, part of a round that values the company at $1.25 trillion, and it hints at where AI is headed.
Read more: OpenAI Receives $11 Billion in New Funding, Now Valued at $1.25 TrillionOpenAI is working with security firm Hacktron to find weak spots in its own systems, which matters if you trust AI tools with your data.
Read more: OpenAI hires a security firm to hunt for holes in its own systemsTwenty-seven Congo residents are suing a Chinese-controlled cobalt mine for $6.3 million over alleged pollution. Cobalt goes into phone and car batteries.
Read more: Congo residents sue Chinese-owned cobalt mine over alleged pollutionElevenMusic, the AI music tool from ElevenLabs, now sends finished tracks to your SoundCloud profile in a few clicks, with a free trial for Artist Pro members.
Read more: ElevenMusic now lets you post your AI-made tracks straight to SoundCloud
