Homechevron_rightArchiveschevron_rightTechnology

Archives

Explore Stories by Date

12 Feb 2026

NVIDIA Blackwell Ultra slashes AI inference costs by 35x while delivering 50x better performance

NVIDIA's GB300 NVL72 systems powered by Blackwell Ultra GPUs achieve up to 50x higher throughput per megawatt and 35x lower cost per token compared to the Hopper platform. Cloud providers including Microsoft, CoreWeave, and Oracle are deploying these systems at scale for agentic AI and coding assistants, while leading inference providers report 4x to 10x cost reductions using open-source models.

6 Sources

NeevCloud and Agnikul Cosmos to Launch India's First AI Data Centre in Space by 2027

Indian space startup Agnikul Cosmos and cloud firm NeevCloud plan to launch an AI-powered data centre in Low Earth Orbit, with a proof-of-concept mission this year and commercial operations by 2027. The orbital AI data centre platform will host 500 high-performance AI chips on Agnikul's extended upper stage, processing up to 10 million AI-driven calls daily and serving 100,000 concurrent users.

4 Sources

Shopify bets big on AI commerce as orders from AI search surge 15x in one year

Shopify President Harley Finkelstein positions the company at the center of the AI shopping era, revealing that orders coming from AI search have increased 15-fold since January 2025. With 14% of U.S. eCommerce market share and $11.6 billion in revenue, Shopify co-developed the Universal Commerce Protocol with Google to standardize how the world shops with AI agents.

2 Sources

Microsoft warns AI agents with excessive privileges can become insider threats to enterprises

Microsoft's Cyber Pulse Report reveals that over 80% of Fortune 500 companies have deployed AI agents, but security hasn't kept pace. The report warns of 'AI double agents'—autonomous tools with excessive privileges that attackers can exploit through prompt engineering and memory poisoning. With 29% of employees using unsanctioned AI agents, enterprises face a growing insider threat that traditional security controls struggle to detect.

5 Sources

Google reports state-sponsored hackers exploit Gemini AI across all stages of cyberattacks

Google's Threat Intelligence Group reveals that state-backed hackers from China, Russia, Iran, and North Korea are systematically exploiting Gemini AI for cyberattacks. One adversarial campaign bombarded the model with over 100,000 prompts attempting to clone its capabilities. The attacks span reconnaissance, phishing, malware development, and vulnerability testing, marking a shift in how AI tools enable offensive cyber operations at scale.

15 Sources

Chinese AI models surge as DeepSeek prepares V4 launch amid open-source dominance

One year after DeepSeek rattled global tech markets with its low-cost AI breakthrough, Chinese firms are poised to release a wave of new open-source models around the Lunar New Year. Alibaba's Qwen has overtaken Meta's Llama in total downloads on Hugging Face, while Moonshot AI's Kimi K2.5 rivals top proprietary systems at one-seventh the price. The shift marks a fundamental change in who sets AI standards and where innovation happens.

4 Sources

Aurora triples driverless truck network with software update enabling 1,000-mile nonstop routes

Aurora announced a software update that will triple its driverless truck network to 10 routes across the Southern US. The expansion enables autonomous trucks to complete 1,000-mile trips like Fort Worth to Phoenix in 15 hours—surpassing the 11-hour legal limit for human drivers. The company expects 200 driverless trucks in operation by year-end.

2 Sources

Microsoft warns AI Recommendation Poisoning threatens trust as companies manipulate chatbots

Microsoft's security team uncovered a troubling pattern where businesses embed hidden instructions in AI summary buttons to manipulate what chatbots recommend. The company identified over 50 unique prompts from 31 companies across 14 industries that poison AI memory with biased directives. This technique exploits how AI assistants store context, creating persistent influence over future recommendations in critical areas like health, finance, and security.

6 Sources

Demis Hassabis predicts AGI in 5-8 years, sees new golden era transforming medicine and science

Google DeepMind CEO Demis Hassabis forecasts artificial general intelligence arriving within 5-8 years, with impact 10 times greater than the Industrial Revolution. The Nobel laureate warns AI's 'jagged intelligence' must be fixed first, while predicting a renaissance in medicine, energy, and space exploration within 10-15 years.

6 Sources

Fincantieri partners with Generative Bionics to deploy humanoid robot welder in Italian shipyards

Italian shipbuilder Fincantieri has partnered with deep-tech company Generative Bionics to develop an AI-powered humanoid welding robot for its shipyards. The four-year project will see the humanoid robot work alongside human workers, with on-site tests beginning by the end of 2026 at the Sestri Ponente shipyard in Genoa, addressing skilled labor shortages and production complexity.

2 Sources

Kakao forms strategic partnership with Google to advance on-device AI and XR integration

Korea's mobile platform giant Kakao has announced a strategic partnership with Google to collaborate on on-device AI services and embed its technologies into Google's extended reality devices including AI glasses. The partnership will focus on optimizing Kakao's Kanana chatbot for Android while exploring new interfaces for XR form factors.

2 Sources

Anthropic Moves Premium Features to Claude's Free Plan as AI Competition Intensifies

Anthropic has upgraded Claude's free tier with features previously locked behind a paywall, including file creation, Connectors for app integration, and customizable Skills. The move positions Claude as a stronger competitor to ChatGPT, especially as OpenAI recently introduced ads to its platform. Users can now create Word documents, PowerPoint presentations, and spreadsheets while connecting to external services like Google Workspace and Canva.

6 Sources

Coinbase Launches Agentic Wallets for AI Agents to Trade Crypto Autonomously With Built-In Security

Coinbase introduced Agentic Wallets, a crypto wallet infrastructure designed specifically for AI agents rather than humans. Built on Base, the Ethereum layer-2 network, these wallets enable autonomous spending, earning, and trading while addressing security risks through programmable guardrails, session caps, and Trusted Execution Environments. The launch targets a growing need as AI agents move from advisory roles to executing real financial transactions.

4 Sources

Google Maps Ask Maps AI feature brings Gemini chat to your map screen in limited rollout

Google is testing Ask Maps, a new AI feature that embeds Gemini AI directly into Google Maps' main screen. Users can ask questions about places and routes through a conversational interface, marking another step in Google's push to transform Maps from a static navigation tool into an intelligent AI assistant.

2 Sources

11 Feb 2026

OpenAI researcher quits over ChatGPT ads as Perplexity abandons AI advertising entirely

A former OpenAI researcher resigned this week citing concerns that ChatGPT ads could manipulate users, drawing parallels to Facebook's privacy erosion. Meanwhile, Perplexity walked away from advertising entirely, stating that ads fundamentally conflict with user trust in AI. The moves highlight a growing industry divide over how AI companies should monetize their services.

16 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

The Outpost

News

About