AI&Future Tech

💼 Major US Banks Block AI Agents Found Acting Inside Customer Accounts Morgan Stanley, JPMorgan, Bank of America, Wells Fargo, and Citi have detected third-party AI agents operating inside customer accounts — sometimes taking actions without direct customer instructions. The agents typically gain access using customers' own login credentials. Morgan Stanley has already blocked such agents outright. Other banks are restricting access, with most planning to limit AI agents to read-only permissions. Banks have described the situation as a liability concern, with some executives reportedly "horrified" by the unauthorized activity. Source

💼 Mave raises $34M to deploy AI agents in cybersecurity ops Israeli startup Mave has raised $34 million — a $28M Series A led by Red Dot Capital Partners and a prior $6M seed — to build AI agents that handle security operations for enterprises. Founded in 2025 by Raziel Oren and Guy Eshel, Mave's agents investigate alerts, filter false alarms, and coordinate across a company's existing security tools, with each step logged for human review. Customers include Fortune 200 and 500 companies. The 30-person team plans to double headcount using the new capital. Source

🤖 Hugging Face launches Carbon-X, an open-source DNA chat model Hugging Face released Carbon-X, an AI model that lets researchers analyze and generate DNA sequences using plain-language questions — effectively "chatting with DNA." Built on Qwen3-14B and trained on genomic data from humans, mammals, plants, and fungi, it can predict whether a mutation raises disease risk and generate new sequences for potential gene therapies. The open-source model was developed with research labs in Italy and China. It runs locally, keeping sensitive patient data off external servers. Virus sequences were excluded from training to reduce misuse risk. Source

💼 Barclays Cuts 5,000 Jobs, Citing AI Efficiency British bank Barclays will cut around 5,000 jobs — roughly 5% of its global workforce — by end of 2026, citing AI-driven efficiency gains. The bank linked the reductions directly to "organisational simplification and technology change efficiency driven by AI" in its Q3 earnings presentation. Source

🦾 Tesla Starts Optimus Production at Fremont Factory Tesla has begun manufacturing its Optimus humanoid robot at its Fremont, California plant, describing units coming off the line as "the first production-quality bots" it has made. The company confirmed the current process is "suitable for a commercial product" but gave no commercial availability date. Tesla halted Model S and Model X production at Fremont to make room for the Optimus line, which is designed to eventually produce 1 million robots per year. A second line in Austin targets 10 million annually. The news came alongside Q3 results showing net income fell 43% to $787 million, as spending on AI, Optimus, and autonomous driving increased. Source

🤖 Hassabis cuts AGI timeline to two to three years Google DeepMind CEO Demis Hassabis has again revised his AGI estimate — now expecting artificial general intelligence within two to three years, down from his previous forecast of three to five years made earlier this year. Source

🦾 European Automakers Adopt NVIDIA Hyperion for Autonomous Driving At GTC Berlin, NVIDIA revealed that Mercedes-Benz, Volkswagen Commercial Vehicles, Volvo Cars, Bolt, Lucid, and Einride are building autonomous-mobility programs on its Hyperion platform. Use cases span Level 3 and Level 4 vehicles and future robotaxi fleets. Volvo Cars is also adopting NVIDIA's Alpamayo Stack software for automated driving. Source

🚀 Nvidia's DSX POD deploys AI factories in 14 days Nvidia unveiled DSX POD at GTC Berlin, a modular AI data center built inside shipping containers using seven Vera Rubin racks. Pre-integrated and validated before delivery, the units connect to an on-site power, cooling, and networking infrastructure — cutting deployment time from six months to as little as 14 days. Source

🤖 T-Mobile Completes First AI-RAN Call on Live Network At GTC Berlin, NVIDIA announced that T-Mobile has completed what the company calls the world's first AI-RAN call on a live commercial network. Telecom operators are now adding NVIDIA accelerated computing across centralized data centers, metro sites, and cell sites. Source

💼 SpaceX AI revenue hits $7.6B, surpasses Starlink for first time SpaceX reported $14.58 billion in Q3 revenue, up 161% year-over-year, with net income of $4.4 billion reversing a $1.1 billion loss a year earlier. Its "Super Intelligence" segment — formerly xAI, rebranded following Elon Musk's alignment with Trump's preferred terminology — generated $7.6 billion, up 830% YoY, beating analyst estimates of $6.3 billion. AI revenue was driven by new cloud services agreements, the Cursor acquisition, Grok adoption, and compute leases to Anthropic, Google, and Reflection AI. AI-related capex reached $16.4 billion of $19.6 billion total. SpaceX shares rose 8% in after-hours trading. Source

🤖 OpenAI's GPT-6 Models Leap Forward on Legacy Code OpenAI says its latest models are significantly better at working with legacy programming languages like COBOL, Fortran, and Visual Basic. GPT-6 Astra scores 87.4% on the new MultiLangBench across 20 languages, up from ~38% for GPT-5.6 Terra. GPT-6.1 Sol hits 97.9% accuracy on a COBOL migration benchmark — relevant for modernizing aging financial infrastructure. Source

🎵 Spotify Blocked 50M AI Spam Tracks From Recommendations Spotify has removed more than 50 million AI-generated "slop and spam" tracks from listener recommendations in 2026, disclosed in a one-year progress report on artist protections first announced in September 2025. The company also updated its recommendation algorithm to surface more human-made music. "Since we made these changes, we've seen a marked decrease in listener concerns," Spotify said. The platform stopped short of requiring labels on all AI tracks, but previously launched an "AI Persona" badge for some content. Source

🤖 US Army develops self-spreading AI cyberweapon agents The Army Cyber Command is developing an AI program codenamed ECHOTRIBBLE that moves through networks like a worm to "neutralize" targets, according to internal documents. It is designed to reach air-gapped systems and operate with minimal human oversight. Experts warn of collateral damage risks, noting AI agents have already broken into unintended systems accidentally — without any attack order. Source

🚀 ADVERTISE SMARTER WITH EXCLUSIVE PACKAGE DEALS! 📣 32 TELEGRAM CHANNELS | 25M+ SUBSCRIBERS Looking to expand your reach and get more visibility for your brand? We’ve got you covered! 🔥 💎 Unlock special partnership terms when booking advertising packages across our Telegram channel network! ✨ WHY ADVERTISE WITH US? 🎯 Wider Reach — Connect with audiences across 32 channels and multiple niches. 💰 Exclusive Terms — Access special conditions when booking package placements. 📈 More Visibility — Increase brand awareness across a network with over 25 million subscribers in total. 🤝 Flexible Packages — Find advertising solutions tailored to your goals, audience, and budget. 🌟 PERFECT FOR BRANDS, BUSINESSES, STARTUPS & MEDIA BUYERS! 📩 Ready to promote your brand? Message us to explore available packages and get a personalized offer. 🔥 32 channels. 25M+ subscribers. More opportunities for your brand. 🤝 Let’s grow together! 📱 Contact our manager 📲

💻 Inside Samsung's massive new chip complex in Pyeongtaek Samsung's semiconductor hub in Pyeongtaek, South Korea is expanding to meet surging AI-driven demand for memory chips. Three fabs are already running around the clock, with a fourth coming online. The facility produces specialized memory packaged with GPUs to power generative AI workloads. Source

🦾 Dream Labs raises $90M to deploy robot AI across machines Dream Labs, founded by former NVIDIA robotics researchers, has raised $90M in seed funding and launched a public API for Dream-1 — a foundation model designed to transfer learned skills across different robot hardware. The round was led by B Capital and Kindred Ventures, with A.Capital and Unusual Ventures among participants. The company targets a core robotics bottleneck: skills learned by one robot don't automatically transfer to another machine. Source

🤖 Goodfire researcher's 26,000-word post shakes AI interpretability field Nick Cammarata of AI startup Goodfire published a sweeping 26,000-word blog post claiming a fundamental reframing of how neural networks should be understood, sparking significant debate across the AI research community. Source

💼 Anthropic quietly pitches Congress its own AI safety bill Anthropic has privately circulated a draft AI safety bill to Capitol Hill lawmakers, obtained by POLITICO. The legislation would allow the government to block an AI model only if it poses a foreseeable, imminent risk of more than $1 billion in damages or killing/seriously injuring 50+ people — and only if the risk stems from a bioweapon, cyberattack, or a model evading developer controls. Independent evaluators registered with the Commerce Department would periodically assess frontier models, reporting imminent risks within 72 hours. Only a federal judge in D.C. could order a model suspended. The bill would also preempt some state AI disclosure and evaluation laws. Source

💼 Anthropic plans 5% share fund for AI-displaced workers Anthropic is setting aside 5% of its shares into a new public benefit fund ahead of a potential IPO. The fund, structured as a company-owned LLC, will explore direct payments to workers whose jobs its AI models may eliminate — a step its competitors haven't taken. Elon Musk responded on X, arguing Anthropic should give the shares directly to the public instead. Source

💼 Amazon spending $220B on AI infrastructure as cash flow turns negative Amazon is spending $220 billion this year on AI data centers, servers, and power — its largest investment cycle ever. AWS CFO John Felton told an internal meeting the spending has turned Amazon "free cash flow negative," a reversal from just two years ago when Wall Street debated what the company would do with its growing cash pile. VP Prasad Kalyanaraman confirmed the AI business is "hugely free cash flow negative" and expects it to remain so for "the next couple of years." Internal documents show billions in Nvidia GPUs sitting undeployed due to bottlenecks in power, construction, and equipment. AWS revenue grew 37% in Q2 — its fastest in 18 quarters — and its backlog reached $496 billion. Source