AI&Future Tech
رفتن به کانال در Telegram
🤖Your daily source for AI, technology and innovation. Breaking news, major launches, research, robotics, Big Tech and the future of intelligent systems. Advertisement: @TGowner999, @topovik_999
نمایش بیشتر1 718
مشترکین
-3324 ساعت
-267 روز
-5930 روز
آرشیو پست ها
1 718
🤖 Claude "Paints" Images Using 700,000 Brushstrokes
Anthropic's Claude Fable 5.2 generates images by writing Python code that simulates brushstrokes on a virtual canvas — complete with bristle texture — rather than computing pixels directly. A single photorealistic parking garage scene required roughly 700,000 strokes, according to Anthropic employee Jake Eaton.
Source
1 718
🤖 Bland releases Tofu-1 speech model for phone calls
Bland has launched Tofu-1, a full-duplex speech-to-speech model designed for noisy, high-stakes phone calls. It is currently in enterprise research preview, with no benchmarks, model size, or pricing disclosed yet.
Source
1 718
🤖 Gartner: AI to create net 1.7M jobs globally by 2030
Gartner projects AI will generate 10.6 million new jobs worldwide between 2025 and 2030 while eliminating 8.9 million — a net gain of roughly 1.7 million. The expected losses would affect less than 2% of workers, though over 435 million people (20%+) would need to adapt how they work.
Over 100,000 AI-related job-cut announcements have already been recorded in the US this year, per Challenger, Gray & Christmas. Goldman Sachs chief economist Jan Hatzius expects unemployment to stay low near-term, with possible upward pressure in following years from aggressive AI adoption — which he called temporary, not a "jobpocalypse."
Source
1 718
🚀 SpaceX catches Starship vehicle with "chopstick" arms for first time
SpaceX successfully caught the Starship vehicle mid-air using the mechanical arms of its "Mechazilla" tower — a first for the upper stage, though the booster catch was aborted. SpaceX had previously caught boosters this way, but never the Starship vehicle itself.
Source
1 718
💼 Anthropic CEO flags "huge uncertainty" ahead of record IPO
Anthropic CEO Dario Amodei has described buying compute as "the deepest business dilemma" his company faces, days before what could be the largest IPO in history. The company's commitments now total nearly $901 billion, with no clear guarantee future revenue will cover those costs. The planned IPO is intended largely as a financial buffer against that uncertainty.
Source
1 718
💻 Apple Silicon Becomes a Quiet Enterprise AI Platform
Meta's latest open-weight model is optimized to run on Mac Studio, and apps like Osaurus now run local AI models on 16 GB Apple silicon laptops — no network connection required. Two models even run on 8 GB devices.
The combination of Apple silicon and the MLX framework is turning Macs into capable local AI servers, reviving a role Apple last filled with the Xserve, discontinued in 2011.
Source
1 718
🦾 DARPA Tests GPS-Free Autonomous War Robots at Fort Knox
DARPA tested four autonomous ground robots at Fort Knox, Kentucky, as part of its RACER program. The uncrewed vehicles navigated without GPS by building terrain maps on the fly using onboard sensors, independently searched for enemy targets, and coordinated with each other or operated solo depending on the scenario.
Human operators remained off-site, assigning tasks and making all engagement decisions. In one demonstration, a robot spotted a target and halted — a human then triggered a simulated rocket strike. DARPA's Jason Gregory said the goal is moving "sensors and effectors forward to the tactical edge while keeping the operator out of harm's way." The five-year program also drew on observations from the war in Ukraine.
Source
1 718
💼 Ex-Anthropic researcher cites IPO slide in calling for AI pause
Former OpenAI and Anthropic researcher Jacob Coxon says an Anthropic IPO roadshow slide showing accelerating algorithmic efficiency was a factor in his departure. He claims Anthropic can now retrain its Mythos model with a fraction of the original compute required. Coxon is calling for a pause on frontier model development, arguing competitors need time to catch up before human-level AI becomes a commodity.
Source
1 718
💼 Amazon plans to index every product sold online
Amazon is building a database called the "World Product Graph" to catalog products sold across the entire internet — not just on its own platform. The goal is to keep shoppers from turning to Google or Reddit for product research, and to position Amazon as the default index for AI shopping agents.
Source
1 718
💻 Microsoft blames failed server health checks for global outage
Microsoft has attributed a recent global outage to failed server health checks rather than a cyberattack. The disruption affected governments, hospitals, and courts worldwide.
Source
1 718
🤖 OpenAI launches Daybreak AI security tools for pen testing and patching
OpenAI has entered the cybersecurity market with two products under its Daybreak brand. Daybreak Red Team uses coordinated AI agents to examine application source code, conduct penetration testing, and verify fixes. Defense Factory investigates security findings and prepares patches. Red Team is currently available through an early-access program.
Source
1 718
💻 Microsoft Outage Enters Day Two, Hitting 74 Countries
Microsoft's global outage, which began Thursday, has stretched into a second day, leaving users unable to access Outlook, Teams, Copilot, SharePoint, and other Microsoft 365 services across 74 countries.
Downdetector recorded over 130,000 US Outlook problem reports at peak. Microsoft attributed the disruption to servers unexpectedly becoming unavailable — not a cyberattack or traffic spike. Recovery has been slow, with some servers going unresponsive again after being brought back online. As of Friday morning, Microsoft reported "additional progress" but warned of continued intermittent issues with no full-recovery timeline given.
The outage has rippled into AI tools: Anthropic and OpenAI both reported disrupted Microsoft 365 connectors, and Claude Code faced authentication issues. Air India and IndiGo rerouted email services; the US House of Representatives and government agencies in Singapore, the Philippines, and Japan also reported impacts.
Source
1 718
💼 Cerebras stock jumps 17% on revenue beat and Microsoft deal
AI chipmaker Cerebras Systems beat Q3 revenue expectations and announced a new partnership with Microsoft, sending its stock up over 17% in late trading. The company also reported a loss of 67 cents per share before certain costs. Cerebras now has cloud deals with Microsoft, Amazon, and Oracle.
Source
1 718
🤖 IBM's Granite Models Cut Prompt Injection Attacks by 47%
IBM released two experimental models — Granite-4.1-3B-LLMON-preview and Granite-4.1-8B-LLMON-preview — trained to separate trusted instructions from untrusted data. On the OPI benchmark, both outperformed the best frontier model IBM tested, blocking 47% more prompt injection attacks. Fine-tuned from Granite 4.1 base checkpoints, both models are available under an Apache 2.0 license.
Source
1 718
💼 Vanta Launches AI Governance and Risk Quantification Tools
Compliance automation firm Vanta unveiled several new products at VantaCon 2026 in San Francisco, including AI Governance — a tool giving security teams a unified view of every AI system across a company. Also launched: Risk Quantification, Autonomous Vendor Assessments, and a Vanta Agent designed to handle entire jobs autonomously. All are part of Vanta's Agentic Trust Platform.
Source
1 718
🦾 SoftBank Leads $500M Round in Agile Robots at $8.5B Valuation
SoftBank has led a $500M Series F round in Munich-based humanoid robotics startup Agile Robots, valuing the company at $8.5B and bringing total invested capital to $2B. NVIDIA, Hitachi, Hitachi Ventures, and T.Capital also participated. The funding will support expanded production of Agile Robots' industrial automation systems.
Source
1 718
🤖 Australia plans mandatory safety audits for frontier AI
Australia's government is drafting legislation that would require frontier AI developers to submit models for independent safety testing. A new regulator would oversee audits and impose penalties for non-compliance. The Albanese government says it intends to establish domestic rules before engaging in international AI governance talks.
Source
1 718
💼 Nvidia Board Meets to Decide Jensen Huang's Fate
Nvidia's board met Wednesday night to decide whether to oust founder and CEO Jensen Huang, Reuters confirmed, after the Wall Street Journal reported the company was preparing to fire him. Nvidia shares fell more than 8% in after-hours trading, erasing over $300 billion in market value.
The crisis follows NYT reports of two extramarital affairs — one involving a woman US officials suspect of spying for China, and a second with an Nvidia executive who reported to Huang. The FBI is also investigating separately from the board's own inquiry.
Bernstein analyst Stacy Rasgon cut his price target from $400 to $200, warning that Nvidia's likely primary beneficiary would be China. Elon Musk said hiring Huang would be "an honor." The board had publicly backed Huang as recently as November 8.
Source
1 718
💼 OpenAI Quarterly Revenue Hits $10B for First Time
OpenAI's quarterly revenue has crossed $10 billion for the first time, up nearly 50% from the $6.7 billion recorded in Q2.
Source
1 718
🤖 Google DeepMind's AlphaGenome 2 maps DNA variants to specific cell types
Google DeepMind has released AlphaGenome 2, an AI model that predicts how DNA variants affect gene regulation in individual cell types and states — not just tissues. Trained on over 100 million single-cell profiles, it targets a key blind spot: a variant may only matter in one specialized or disease-activated cell, making it invisible to broader genomic models.
AlphaGenome 2 is available now for non-commercial research and in private preview for commercial use.
Source
