Meta scans your height and bones with AI Meta is rolling out an AI age-detection system that reads photos and videos for height and bone structure to estimate whether a user is under 13. The company says it does not identify specific people, only estimates general age. The details: Visual layer: AI reads photos for height and bone structure. Text layer: Same model scans bios, captions, comments and DMs for grade or birthday cues.
Legal backdrop: Comes weeks after a New Mexico jury hit Meta with $375M for misleading parents about platform safety. Meta's framing: "This is not facial recognition." The model reads "general themes and visual cues." Why it matters : Bone structure and height are how a doctor guesses age without an ID and Meta is now running that estimate on every photo uploaded to Instagram and Facebook. Privacy-conscious users will wonder if this is really about age verification or whether the same model ends up repurposed for ad targeting or behavioral profiling.
Your data speaks. Does your AI listen? Enterprises are drowning in data. Yet most AI just sits on top, guessing at context it was never built to understand.
Sema4.ai ’s semantic layer gives AI agents real business context, relationships and governance. So they can reason, not just retrieve. AI intelligence is woven into the very fabric of the platform, top to bottom.
Desk, Invoice Reconciliation, Receivables Matching – all running autonomously, on data they truly understand. Learn More Top AI Courses of the Week AI is reshaping how software gets built. Discover these four GenAI Academy courses to help you code faster, ship reliably and take real projects to production. Vibe Coding with Python Build real Python apps with AI, no experience needed.
Last call, 20% off. From Demo to Production From prototype to deployed product. Early bird, 20% off. Claude Code Mastery Build a production-ready AI coding workflow.
Ship Reliable Software Faster with AI Turn product specs into working code. Apple is paying back iPhone owners $250M for lying about Apple Intelligence Apple agreed to pay $25 to $95 to iPhone 15 and iPhone 16 buyers in the US who purchased between June 2024 and March 2025. Total payout is $250M, settling a class action accusing Apple of false advertising on Apple Intelligence.
Apple shipped a smaller subset under the same brand. Apple's spin: Denied wrongdoing. Called the suit a fight over "two additional features." Leadership: Outgoing CEO Tim Cook caught years of criticism over Apple's AI pace. John Ternus takes over in September.
Why it matters: Apple just paid cash to customers over an AI keynote demo that never shipped. The settlement covers Siri claims and sets a precedent for every product company tempted to stage a slick AI unveil without working software behind it. Anthropic just shipped ten plug-in Claude agents for finance work Anthropic released ten agent templates aimed at financial services firms, covering pitchbooks, KYC screening, valuation reviews, month-end close and earnings analysis. Each ships as a Claude plugin in Cowork or Code, or as a cookbook for Managed Agents on the Claude Platform.
The roster: Pitch builder, meeting preparer, earnings reviewer, model builder, market researcher, valuation reviewer, GL reconciler, month-end closer, statement auditor and KYC screener. Microsoft layer: Claude add-ins for Excel, PowerPoint, Word and Outlook (coming soon) carry context across apps. Data layer: New connectors for Dun & Bradstreet, Verisk, IBISWorld, Guidepoint and Fiscal AI. Moody's MCP: A custom Moody's app inside Claude surfaces credit ratings on 600M+ companies. Why it matters: A pitchbook usually requires three analysts, a comp sheet, a methodology doc and a deck cycle that eats two weeks.
Claude is offering the whole thing as a single template that ships in days. These agents are the inventory yesterday's $1.5B consulting JV was stood up to deploy and the back office of every Wall Street firm is the prize. The EU just turned a decade of disaster news into AI knowledge graphs The Joint Research Centre published an open dataset that uses LLMs to turn global news coverage of disasters into structured storylines and knowledge graphs. It covers more than 3,000 events across 175 countries between 2014 and 2024. The details: How it works: RAG pulls disaster articles from the Europe Media Monitor, then LLMs distill each event into a storyline.
What it captures: Cascading effects databases miss, like flooding that triggers transport disruption, crop loss and disease outbreaks. Coverage: 26 disaster types covering around 80% of global economic losses in EM-DAT over the same window. Bias correction: Pulls from broader sources to surface slow crises like droughts in vulnerable regions. Why it matters: Crisis response runs on patchy databases that rarely talk to each other and a free LLM-built knowledge graph linking cause, effect and response at event level is the missing layer for emergency planners.
Any government or insurer can stand up the same pipeline using public news. This is one of the cleaner examples of useful, non-hyped AI shipping in Brussels. Tool of the day: kickresume The formatting's a mess, you're guessing on every bullet point, and you have no clue if it'll survive the ATS scan. Kickresume fixes all of that: 40+ templates designed by actual typographers An AI writer running on GPT-4.1 that builds your whole experience section in seconds And an ATS checker showing you what's working or failing Throw in 1,500 real examples from people hired at Google and Apple, and you've actually got something that works. 8 million people use it.
There's a free version. See for yourself. Light Bytes Anthropic is building Orbit , a proactive briefing tool inside Claude Cowork that pulls insights from Gmail, Slack, GitHub and Figma. Perplexity Computer is now live in Microsoft Teams, letting users run research, build dashboards and draft documents with ease.
GPT-5.5 Instant is now the default in ChatGPT, jumping from 65.4 to 81.2 on the AIME math test and pulling context from past chats, files and Gmail. OpenAI president Greg Brockman testified he thought Elon Musk was about to hit him during a 2017 meeting where Musk demanded more control over the company. Coinbase cut 14% of its workforce , roughly 693 jobs, with CEO Brian Armstrong calling it a move to an "AI-native operating model" targeting half of all code written by AI.