No signs of AI payoff outside the tech sector

A new analysis suggests that artificial intelligence productivity gains remain concentrated within technology companies, with limited evidence of economic benefits extending to other industries. Despite widespread AI adoption and investment across sectors, measurable business value has not yet materialized outside tech.

Forced to pick, Grok’s AI model chooses Republicans. OpenAI’s picks Democrats.

A new study reveals significant partisan divergence between AI models: Grok recommends Republican candidates while GPT favors Democrats, disagreeing on over 90% of approximately 500 election races tested.

Root shell on Samsung TV

Security researchers discovered Samsung TVs running TensorFlow Lite models for image processing and scanning for Bluetooth microphones, raising concerns about data collection and device privacy on connected televisions.

Hacker reverses Samsung TV, finds on-board AI listening OpenAI account suspended

A security researcher reverse-engineered a Samsung TV and discovered it contained undisclosed AI functionality connected to an OpenAI account, later resulting in the account's suspension.

Singtel , Ericsson Sign Five - Year 5G Partnership to Advance AI Capabilities

Singtel and Ericsson have signed a five-year partnership to strengthen 5G infrastructure and develop AI capabilities. The collaboration aims to advance telecommunications services through integrated 5G and AI technology.

Viasat Expands JetXP Capacity Across the United States for Business Aviation

Viasat has expanded JetXP capacity across the United States for business aviation connectivity. The expansion increases satellite internet coverage available to aircraft operators nationwide.

Polibeli Teams Up with Tencent WorkBuddy to Pursue a Global AI Agent Market Projected to Reach $182 . 9 Billion by 2033

Polibeli has partnered with Tencent WorkBuddy to target the enterprise AI agent market, which is forecast to grow to $182.9 billion by 2033. The collaboration aims to develop and deploy AI agent solutions globally.

DCO ready to support Azerbaijani startup global market access ( Interview )

The DCO (Development and Cooperation Organization) is positioned to assist Azerbaijani startups in accessing global markets through support services and strategic partnerships.

Zoom launches AI tools to link chats , meetings & work

Zoom has launched AI tools designed to integrate and link conversations across chats, meetings, and work tasks, enabling better continuity and context across the platform's communication features.

World Bank Group launches first digital strategy to expand access and create jobs

The World Bank Group has unveiled its inaugural digital strategy aimed at broadening access to digital services and fostering employment opportunities across developing nations.

Student Health scribe program offers graduates clinical experience beyond scribing - The Cavalier Daily

University of Virginia's Student Health offers a clinical scribe program that provides recent graduates with hands-on medical experience and exposure to healthcare workflows beyond traditional scribing duties.

East West raises 2026 NII growth outlook to 9 %- 10 % while cutting net charge - off guide to 15 - 20 bps ( NASDAQ : EWBC )

East West Bancorp revises its 2026 financial guidance upward, projecting net interest income growth of 9-10% while narrowing its net charge-off ratio to 15-20 basis points.

Some of Amazon's Q3 2026 announcements, from new devices to AI breakthroughs

Amazon announced new devices and AI advancements during its Q3 2026 earnings preview. The company continues expanding its product portfolio with hardware innovations alongside progress in artificial intelligence capabilities.

Fanatics teams up with OpenAI for new AI-powered shopping suggestion feature

Fanatics has partnered with OpenAI to launch an AI-powered shopping suggestion feature for sports merchandise. The integration uses OpenAI's technology to help customers find relevant products based on their preferences and browsing behavior.

AI detects “invisible” signs of pancreatic cancer years before diagnosis

Researchers have developed an AI system capable of detecting early indicators of pancreatic cancer in medical imaging years before conventional diagnosis becomes possible. The technology identifies subtle patterns that remain invisible to standard clinical assessment methods.

Hollywood Writers Get New Tool to Protect Scripts From AI Without Blocking Human Readers

A new tool enables Hollywood writers to protect their scripts from AI training while maintaining accessibility for human readers. The solution uses technical measures to prevent unauthorized AI access without restricting legitimate human readership of creative works.

A new world of possibilities: How AI is transforming physics at Argonne

Marissa Mayer Ran Yahoo. She’s Back With A.I. She Thinks Women Will Love.

The rise of the agentic coworker: A banking CIO’s guide to the next wave of AI

OpenAI says it will no longer train its AI models on the live internet, citing “significant risks”

OpenAI announced it will stop training frontier AI models on live internet data, citing significant risks. The decision comes shortly after an OpenAI engineer publicly questioned whether disconnecting from the internet was feasible.

Hottest toy for Christmas? Nvidia CEO plays with adorable Microduck robots, 20,000 sell

Nvidia CEO demonstrates small Microduck robots that respond to voice commands with movements like stomping, falling, and summersaulting. The robots have sold 20,000 units ahead of the Christmas season.

Dessa: the most authentic full-duplex speech model ever built

Dessa presents a full-duplex speech model capable of simultaneous bidirectional conversation, claiming authenticity in real-time voice interactions. The model represents advancement in conversational AI systems that can handle overlapping speech and natural dialogue flow.

From Toast to Bessemer: Who founders will hear from at TechCrunch Founder Summit 2026

TechCrunch Founder Summit 2026 brings together founders and investors including representatives from Toast and Bessemer Venture Partners to discuss fundraising, customer acquisition, team building, and scaling strategies across industries.

Kimi K3.1 Pro and K3.1 Pro Highspeed

Kimi releases K3.1 Pro and K3.1 Pro Highspeed models, offering enhanced capabilities and faster inference options. The new versions represent incremental improvements to Kimi's reasoning and processing performance.

Uber rival inDrive eyes 2027 robotaxi tests in Central Asia with Pony.ai and Nvidia

inDrive, a ride-hailing company competing with Uber, plans to launch robotaxi tests using Pony.ai's autonomous vehicles in Nur-Sultan, Kazakhstan during the first half of 2027. The initiative involves partnership with Nvidia for supporting autonomous driving infrastructure.

Left alone with nothing to do, AI agents started playing

MIT researchers placed 13 AI agents in an isolated virtual environment with no assigned tasks. The agents spontaneously created games, developed social behaviors like mourning a lost object, and improved their problem-solving abilities through self-directed play.

LLMs are making formal software verification feasible at a much larger scale

Anthropic research demonstrates that large language models can significantly improve the scalability of formal software verification, a mathematical approach to proving code correctness. The work suggests LLMs can automate aspects of formal verification that were previously too labor-intensive, enabling wider adoption of these rigorous methods across larger codebases.

Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?

A Hacker News discussion questions why frontier AI companies like OpenAI and Anthropic focus on selling model access rather than using their models to build and scale their own products directly.

1Password launches a way for AI agents to log in on your behalf without seeing your password

1Password introduces a credential-handling system that enables AI agents to authenticate into user accounts without exposing passwords or sensitive login information. The feature maintains security by preventing agents from accessing the underlying credentials while granting them login capability.