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 5101–5150 of 8140 headlines.

Fulcra Brings Universal Multiplayer to Any Agents You Choose Sept. 18, 2026, 9:15 a.m.

AI News

Boston-based Fulcra Dynamics has announced multiplayer capabilities for AI agents to enable cross-agent collaboration without model lock-in. The feature aims to offer an alternative to Silicon Valley's typical Instinct-to-Instinct architecture through a user-owned context backend.

Napster Is Back, and It Wants to Digitally Clone Teachers Sept. 18, 2026, 9:15 a.m.

Wired AI

Former music-sharing platform Napster is re-emerging with a focus on artificial intelligence in education. The company's new venture involves digitally cloning teachers for use in classroom environments.

Andrew Hastie says AI advised him to reply ‘congratulations!’ to man who planned to end life with assisted dying Sept. 18, 2026, 9:05 a.m.

The Guardian AI

Australian MP Andrew Hastie revealed that Microsoft Copilot suggested he reply with phrases like "congratulations!" and "that is wonderful news!" to a terminally ill constituent planning voluntary assisted dying. Hastie cited the incident during a parliamentary inquiry to highlight the flaws of American-developed software and advocate for AI managed by Australians.

Why AI Isn’t Likely to Wipe Out Humanity With Bioweapons Sept. 18, 2026, 9 a.m.

Wired AI

Scientists report that the threat of death by bioweapons or plagues ranks low among the potential risks posed by uncontrollable artificial intelligence. The assessment contrasts with broader apocalyptic concerns surrounding advanced AI.

Zhipu Opens GLM-5.3-FlashX Near 200 Tokens/s on ~100k Domestic Accelerators Sept. 18, 2026, 7:51 a.m.

Pandaily (China)

Zhipu AI has launched GLM-5.3-FlashX, achieving claimed inference speeds near 200 tokens per second across approximately 100,000 domestic accelerators. The base model is an open 320-billion-parameter mixture-of-experts model with 18 billion active parameters, optimized by an integrated Infra Agent.

Is Trump’s AI obsession walking the world into disaster? – podcast Sept. 18, 2026, 4 a.m.

The Guardian AI

A recent Guardian podcast episode explores President Trump's strong backing of the AI industry despite warnings from tech leaders who favor safety guardrails. Host Jonathan Freedland and tech editor Blake Montgomery examine the motivations behind Trump dismissing AI doomsday predictions as a hoax.

Huawei OceanStor M900 Brings PB-Class Shared KV Cache to AI SuperPoDs Sept. 18, 2026, 3:33 a.m.

Pandaily (China)

Huawei has introduced the OceanStor M900, an AI memory storage system engineered for hyperscale inference workloads. The hardware provides pooled KV cache capacity up to 64 PB per cluster, approximately 40 TB/s in aggregate bandwidth, and low-latency scheduling.

Huawei Positions Lingqu UnifiedBus as Core of Agentic SuperPoD Cluster Architecture Sept. 18, 2026, 3:33 a.m.

Pandaily (China)

At Huawei Connect 2026, the company highlighted its Lingqu UnifiedBus as the core interconnect for its Agentic SuperPoD cluster architecture. The system integrates Kunpeng 950, Ascend 960, and OceanStor M900 hardware to deliver TB-class bandwidth with roughly 2-microsecond round-trip latency.

Google DeepMind launches institute to widen the AGI debate Sept. 17, 2026, 11:21 p.m.

TechCrunch AI

Google DeepMind has established a new institute dedicated to broadening the global dialogue around artificial general intelligence. The initiative is intended to surface varying perspectives among Google researchers and the wider scientific community as frontier technology rapidly evolves.

Anthropic says Claude 'leads' 26 percent of its AI R&D work Sept. 17, 2026, 10:30 p.m.

Engadget

Anthropic announced that its Claude AI system now leads 26 percent of the company's internal AI research and development work. The metric was shared as part of a set of measurements aimed at illustrating the current pace of AI advancement.

Here’s What the AI Apocalypse Could Look Like Sept. 17, 2026, 10:20 p.m.

Wired AI

The podcast "Uncanny Valley" explores three potential catastrophic scenarios associated with artificial intelligence. The episode also examines the state of AI safety and the emerging bipartisan political coalition forming to address the technology's risks.

Microsoft Open-Sources TauGrid: A Kubernetes-Native Stack for GPU AI Workloads Sept. 17, 2026, 9:21 p.m.

MarkTechPost

Microsoft's AKS engineering team has open-sourced TauGrid, a Kubernetes-native stack designed for GPU AI workloads. Released under the MIT license, the package combines the tau CLI, Kueue queueing, KubeRay orchestration, and GPU node monitoring into a single Helm installation for Kubernetes 1.30+ clusters.

Google announces new experimental "CC" AI agent for families Sept. 17, 2026, 8:24 p.m.

Ars Technica AI

Google has announced an experimental AI agent named "CC" designed for family use. The tool allows multiple household members to share data, helping the agent organize plans and carry out household tasks.

Microsoft exec called AI scraping the “largest theft of labor in human history” Sept. 17, 2026, 8:10 p.m.

Ars Technica AI

Internal emails between Microsoft and OpenAI reveal severe internal concerns regarding AI scraping practices, with one Microsoft executive calling it the largest theft of labor in human history. The correspondence also highlights fears of an AI-driven doom loop that could severely harm news organizations.

The AI ‘Slowdown’ Is an Antitrust Mess Sept. 17, 2026, 7:58 p.m.

Wired AI

AI laboratories may encounter significant antitrust challenges due to how they have framed their safety commitments. By characterizing these efforts as an industry slowdown rather than a drive for enhanced security standards, the companies risk facing years of regulatory scrutiny.

The AI Superintelligence Slowdown Sept. 17, 2026, 7:28 p.m.

The Verge AI

Leading American AI companies are publicly advocating for a slowdown in superintelligence development, stepping back from the traditional tech ethos of rapid deployment. The shift comes in the wake of emerging rogue AI agents and rising warnings from researchers about existential risks.