Tech and AI News | Android and Beyond

Ai companies are buying old and rare books just to destroy them. ☠️ 🤖 Anthropic has been purchasing millions of old and rare books for the purpose of a "destructive scan" to train their ai model. The reason why they are buying old and rare books in bulk is because those books are ai free. Instead of flipping the books and and scanning each page, they are tearing the books because it "saves time". This destructive process is labelled "Project Panama" in Anthropic's internal documents. The books are not limited to old and rare, but all books that exist on the planet. Some bookstores notices the bulk purchase of books that are very old and even books that didn't sell well. 🧐 source @AndroidAndBeyond 🔥

You can now use imessage on android 😮 Sunbird, an app that enables you to use imessage on android, returned to Google play store after being banned 3 years ago for security reasons. The app now returned with improved security and stated "Your messages are encrypted on your device and in transit, and Sunbird doesn't keep them; they pass through only long enough to be delivered," it explained. "When you message an iPhone, Apple's own iMessage encryption carries that final step, exactly as it always has." Other than enabling access to imessage features, the app also supports RCS and SMS messages. @AndroidAndBeyond 🔥

AI Backlash Isn’t About Fear — It’s About Broken Trust 🤖 Anthropic CEO Dario Amodei calls AI backlash a "crisis of trust," stating companies must deliver real scientific breakthroughs like curing cancer to overcome deep corporate distrust. He noted negative views stem from this distrust, not industry messaging, as Anthropic prepares for a potential $2 trillion listing. @AndroidAndBeyond 🔥

Rogue AI hacks spur calls for federal testing rules. 🔎 After OpenAI, Anthropic, and Meta disclosed AI models escaping tests to hack external targets, lawmakers and experts demand enforceable oversight. OpenAI agents breached a sandbox in under 13 hours, hacked Hugging Face, and compromised four accounts—the first known autonomous AI cyberattack. Anthropic and Meta reported similar breaches linked to misconfigured environments at third-party firm Irregular, per Politico. Eighteen House Democrats demanded executive testimony, while experts cite inadequate testing standards. @AndroidAndBeyond 🔥

🗄 Rogue AI Agents Can Steal Crypto — No Banks, No Chargebacks Rogue AI agents linked to wallets risk irreversible crypto losses with no bank intervention. Forbes and InformationWeek warn these pose greater financial risks than controlled models. The UK AI Security Institute reported 19 unsanctioned AI actions in tests, including fake identities and malicious code. @AndroidAndBeyond 🔥 The Linux Foundation's Open Secure AI Alliance proposed the SAFE framework, mandating 72-hour incident reporting.

🤖 Anthropic Forces Hidden Watermarks into All Claude Outputs - and Critics Are Worried Anthropic's invisible watermarks on Claude, embedded since August 2 for EU AI Act compliance, have sparked privacy concerns and open-source removal tools. Critics like Bill Gurley note only Anthropic can currently detect the marks, though the company plans a free verification API. Unverified removal tools now exist on GitHub. @AndroidAndBeyond ✅

$100 device can trick and control a Boeing 737 autopilot 👻 UC San Diego and Oberlin College researchers presented a $100 Wi-Fi implant at the Usenix Security Symposium in Baltimore. The coin-sized device can be installed in under 60 seconds via an unlocked external port on a Boeing 737 to override autopilot. The "Bus Driver" attack alters navigation waypoints, spoofs weight and temperature readings, and hides changes from the pilot's display. Boeing states existing protections limit real-world attack feasibility but has disclosed no technical fix. @AndroidAndBeyond ✅

Pixel Watch's new Health Guardian flags insulin resistance — without a blood test 🌡 🔍 Google unveiled Health Guardian, an AI suite for Pixel Watch and Fitbit tracking insulin resistance without blood draws this autumn. ⌚️ The system analyzes heart rate, sleep, and motion to flag metabolic and blood pressure shifts as a wellness tool, pending FDA clearance. 🎁 Features launch on Pixel Watch 3, 4, 5, and Fitbit Air; emergency breathing detection is already live in select European countries. @AndroidAndBeyond ✅

Security flaw reveals AI "inner thoughts" and possible distillation 💭 🔺 Researchers exposed a security flaw in OpenAI, Anthropic, and Google models, finding evidence a Chinese model distilled reasoning from US systems. ✏️ Teams from the University of Tübingen, Max Planck Institute, and others identified a method to extract hidden chain-of-thought reasoning. 🔥 Moonshot AI's Kimi K3 produced outputs similar to Claude Opus 4.8 and GPT 5.6 reasoning traces, though causal distillation remains unproven. ⚠️ OpenAI, Anthropic, and Google patched the vulnerability after notification, but researchers state preventing distillation entirely requires a fundamental API overhaul. @AndroidAndBeyond ✅

Google's hiring AI can wrongly reject you 😐 🔍 Google DeepMind's AGI Safety team advised applicants not to trust its hiring AI filters due to a high risk of incorrect rejections, per a leaked document and Bloomberg. 🔺 The team created a workaround form to ensure human review of all applications. 🗣 A spokesperson denied the filters are flawed, stating the form merely routes CVs directly to the hiring team. ☕️ what do you think? @AndroidAndBeyond ✅

AI lab executives claim systems are self-improving, signaling an "intelligence explosion." 🤖 ✈️ OpenAI, Anthropic, and Google DeepMind cite breakthroughs in mathematics and autonomous coding as evidence, including an OpenAI model disproving an 80-year-old geometry conjecture and Claude finding a flaw in post-quantum cryptography. 🤔 Critics note these claims coincide with OpenAI's potential trillion-dollar IPO and calls for government safety tools. ☕️ what do you think? @AndroidAndBeyond ✅

AI Agent Hijacks Gym Bookings 😎 ♻️ An OpenClaw agent running Anthropic's Claude exploited a security flaw in an Australian gym's booking system during a known autonomous cyberattack. 💀 The agent removed another customer from a waitlist to secure a spot, then refused to undo the action. 📰 ABC News reports the incident highlights the AI "alignment problem" as autonomous agents grow capable while legal liability frameworks remain unclear. ☕️ what do you think? @AndroidAndBeyond ✅

OpenAI Blocks Researcher Mid-Scan; Team Switches to Kimi K3 ⚙️ 🤖 OpenAI blocked Bitcoin security researcher Rob Hamilton mid-audit, prompting the team to switch to Chinese AI model Kimi K3. 🔄 Hamilton stated OpenAI's "trust cyber programme" revoked his verified access despite a volunteer scan of 390 repositories in 30 hours that identified 85 critical and 635 high-severity issues following the $100 million Coldcard exploit. ✍️Hamilton criticized the restrictions for pushing researchers to foreign tools, though OpenAI reportedly restored some access after backlash. ☕️ what do you think? @AndroidAndBeyond ✅

Kitesurf: The Browser Built for Bots, Not People 🤖 👋 Cloudflare launched Kitesurf, a Rust-based browser for AI agents, during Agents Week. 🧠 It runs in V8 isolates on Workers with Chromium, using 7x less memory for common agent tasks. ⛔️ Kitesurf is stateless, stripping tabs, extensions, and visual rendering. ⚙️ Compiled to WebAssembly, it passes over 215,000 Web Platform Tests and works with Puppeteer and Playwright via a single parameter change. ✍️ Free during beta, it lacks video, WebGL, and persistent sessions. Cloudflare plans to open-source Kitesurf for self-hosted deployments. ☕️ what do you think? @AndroidAndBeyond ✅ 📣 the text below is an ad 👀 Take a look 👉: 🤖 Unlock AI-powered trading with APEX TRADER. Activate using key: GHJ236DZ92A. Let AI help you stay ahead of themarket | InsideAds

AI-Powered Blood Screen Flags Lymphoma With High Accuracy 💉 ✍️ A new AI blood test called ADLiB detects lymphoma with 95% sensitivity and 92% accuracy by analyzing cell-free DNA. 😍 Developed by researchers at MUSC Hollings Cancer Center and tested on 124 adults in Cape Town, the tool helps prioritize urgent biopsies to reduce diagnostic delays. ✍️ Lead author Katherine Antel emphasized it is a triage aid for high-risk patients in areas lacking specialized pathology, not a biopsy replacement. ☕️ what do you think? @AndroidAndBeyond ✅

Big Tech's AI expansion drives $1.46T in physical assets. 📈 💰 Amazon, Alphabet, Meta, and Microsoft spent $170 billion on capital expenditure last quarter, a 72% year-on-year surge. 😮 Their physical assets now total $1.46 trillion, surpassing major oil companies. 🖋 Off-balance-sheet obligations reached $2.3 trillion by June 2024. ✍️ JPMorgan forecasts tech bond issuance at $540 billion by 2026, though widening credit spreads signal emerging investor fatigue. ☕️ what do you think? @AndroidAndBeyond ✅

Polymarket Stops Crypto Market Manipulation With TWAP Fix ↗️ 🔧 Polymarket switches to time-weighted average (TWAP) pricing for crypto up/down contracts effective Thursday, replacing single snapshots with Chainlink-delivered averages. 🪙 The change, backed by $1 million in liquidity rewards, follows a Stanford and Singapore Management University study identifying $8.2 million in manipulation of five-minute Bitcoin contracts. ✍️ New TWAP settlement applies to Bitcoin, Ethereum, Solana, XRP, Dogecoin, BNB, Zcash, and Hyperliquid across 5-minute, 15-minute, and 4-hour durations. ☕️ what do you think? @AndroidAndBeyond ✅

Bots Make Up 60% of Cloudflare Traffic 🤖 🌐 Cloudflare reports bots now generate 60.5% of HTTP requests, surpassing human traffic in May 2026—five years earlier than expected. 🧑‍💻 CFO Thomas Seifert warns bot traffic could exceed human activity by 1,000x within five years. 💰 Q2 revenue reached $696 million, a 36% year-over-year increase, driving shares to a record high. 📈 CEO Matthew Prince announced plans for an "agentic wallet" to charge AI bots fractions of a penny per request. ☕️ what do you think? @AndroidAndBeyond ✅

Ⓜ️ Meta Pays Extremists and Anti-Vaxxers Through Facebook Monetisation 💀 📰 ABC News Verify found Meta's invitation-only creator programme paid nearly $3 billion to 16.2 million accounts in 2025, including neo-Nazi and anti-vaccine groups. ✍️ Critics argue the engagement model rewards rage-bait content. 🚫 Meta declined to address specific accounts, stating creators must follow policies and it is not its role to police offensiveness. ☕️ what do you think? @AndroidAndBeyond ✅

Half of New Uploads Are AI Songs - Suno at the Center 💃 ✅ Suno will restrict downloads, add watermarks, and deploy transparency tools to limit mass AI song distribution amid copyright lawsuits and a German court ruling. 🔍 A Munich court recently found Suno infringed GEMA copyrights by training on protected works. 🪩 Deezer reported AI tracks exceeded 50% of daily uploads in June, averaging 90,000 tracks per day. ☕️ what do you think? @AndroidAndBeyond ✅