Nicheloom

The opportunity tracker for new startups.

AI News

Real headlines from AI-dedicated outlets (TechCrunch AI, The Verge AI, OpenAI News, Hugging Face, and more) plus AI-relevant Hacker News stories, pulled every 6 hours — not another analysis, just the fastest real feed of what's happening.

The Sources/Vertical/Function filters above don't apply here — those filter launches, and a news headline isn't a launch. Use Source below to narrow by news outlet instead.

Source

Headline volume, last 14 days

Showing 4151–4200 of 7905 headlines.

Nvidia boss says there is ‘0% chance’ AI destroys the world by 2030 Sept. 21, 2026, 10:31 a.m.

The Guardian AI

Nvidia CEO Jensen Huang stated there is a zero percent chance artificial intelligence will cause the extinction of humanity by 2030. He dismissed recent warnings of superhuman AI ending the human race as irresponsible and overblown doomsday narratives.

UN says AI safeguards can’t wait for certainty Sept. 21, 2026, 10:18 a.m.

The Verge AI

A United Nations scientific panel has warned that governments must regulate capable AI agents before their risks are fully understood. The alert was issued in the UN's first major assessment of the OpenAI hack of Hugging Face, placing AI safeguards firmly on the global diplomatic agenda.

Building standards for the next phase of AI Sept. 21, 2026, 10 a.m.

OpenAI News

OpenAI has outlined a framework for establishing shared global AI standards. The proposal advocates for coordinated evaluation, reporting, and governance mechanisms to strengthen overall AI safety.

Adlyse wants AI agents managing ad campaigns 24/7 Sept. 21, 2026, 10 a.m.

Tech.eu (Europe)

Adlyse aims to deploy AI agents to manage advertising campaigns around the clock. The company is targeting operational tasks in the advertising sector, which has not adopted AI assistance at the same pace as product engineering.

Nvidia’s free AI model could push the UAE closer to the U.S. Sept. 21, 2026, 10 a.m.

Rest of World

While the United Arab Emirates possesses substantial computing hardware, it has largely rented the underlying artificial intelligence. Nvidia's upcoming free AI model could alter that dynamic and potentially align the UAE closer to the United States.

The US has proposed an AI incident alert system to China Sept. 21, 2026, 9:48 a.m.

The Next Web

During talks in Manhattan, US Treasury Secretary Scott Bessent proposed an incident alert system to Chinese Vice Premier He Lifeng. Under the proposal, Washington and Beijing would notify each other of serious AI incidents that qualify as national security threats.

Amazon blocks Meta’s Muse AI agent Sept. 21, 2026, 9:21 a.m.

The Verge AI

Amazon is blocking Meta's Muse AI agent from accessing its platform to shop on behalf of users. The company began displaying messages warning that unauthorized AI agents violate its conditions of use, with reports noting Meta did not notify Amazon prior to deployment.

The White House will not say what Trump’s AI Force is Sept. 21, 2026, 8:47 a.m.

The Next Web

The White House has not clarified whether Donald Trump's proposed 'AI Force' is intended to be a military branch, a regulatory body, or another type of entity. The initiative was announced over the weekend on Truth Social alongside plans for an AI czar, rather than through an executive order.

Radiant D35 RISC-V Dataflow Chip Partners With PrimeBot for Embodied Edge Sept. 21, 2026, 8:22 a.m.

Pandaily (China)

Suzhou-based chip designer Ricore has partnered its Radiant D35 RISC-V dataflow AI chip with PrimeBot's compute ecosystem. The collaboration focuses on enabling edge inference specifically for humanoid and embodied robots rather than traditional cloud training.

CIX Technology Launches AGX X2 Agentic Edge Platform at 80–300 TOPS Sept. 21, 2026, 8:19 a.m.

Pandaily (China)

CIX Technology has introduced the AGX X2 agentic edge platform, which delivers compute performance ranging from 80 to 300 TOPS alongside 64 to 160GB of LPDDR5 unified memory. The platform is designed for native inference on 7B to 122B parameter models and claims to reduce energy consumption per token by more than 50%.

How to Improve Visibility Across Your Enterprise AI Ecosystem Sept. 21, 2026, 7:50 a.m.

AI News

Rapid enterprise AI adoption is outpacing governance, leaving organizations exposed because traditional monitoring tools cannot adequately track AI activity. As comprehensive visibility becomes essential for AI security controls, security teams must implement modern strategies to identify and manage these emerging risks.

Expanding OpenAI Academy with new learning paths Sept. 21, 2026, 7 a.m.

OpenAI News

OpenAI Academy is expanding its offerings with new learning paths tailored for employees, developers, leaders, educators, and students. These programs are designed to help participants develop and demonstrate practical AI skills.

Metris secures $5M to build a unified data layer for energy assets Sept. 21, 2026, 7 a.m.

Tech.eu (Europe)

Metris Energy has raised $5 million in a seed funding round led by PT1 Ventures, Octopus Ventures, AENU, and Blackfinch Ventures, with additional backing from Plug and Play. The company will use the capital to develop its AI-native unified data platform for managing energy assets.

StepFun Launches Step 5 Preview: A 600B-Total, 27B-Active MoE Model With 1M Context for Long-Horizon Agentic Work Sept. 21, 2026, 6:37 a.m.

MarkTechPost

StepFun has released Step 5 Preview, a sparse Mixture-of-Experts model featuring 600 billion total parameters, 27 billion active parameters per token, and a 1-million-token context window. The multimodal model accepts text, image, and video inputs to assist with long-horizon agentic workflows in software engineering, knowledge work, and finance. API access is currently available, with open weights planned for release on October 15, 2026.

Zhipu open-sources ZCode after data dispute and plans no-retention controls for MaaS Sept. 21, 2026, 5:58 a.m.

TechNode (China)

Following an apology and remediation over a data-handling dispute, Zhipu has open-sourced its ZCode repository on GitHub. ZCode is structured as an AI coding workbench featuring desktop, browser, and terminal-agent components. Zhipu also announced that its Model-as-a-Service platform will soon introduce controls allowing users to request no data retention.