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 5901–5950 of 8555 headlines.

Alibaba Cloud Plans to More Than Double Global Modular Data Center Capacity Aug. 11, 2026, 5:31 a.m.

TechNode (China)

Alibaba Cloud plans to more than double its global modular data center capacity in 2026 to meet growing AI computing demands. The company stated that its standardized modular approach enables the deployment of large AI data centers in approximately 100 days.

Li Auto Expects Second-Generation AI Glasses in H2 2027 Aug. 11, 2026, 5:28 a.m.

TechNode (China)

Li Auto anticipates releasing the second generation of its Livis AI glasses in the second half of 2027. The company is evaluating lens-based display technologies with Zeiss and working on native compatibility with HarmonyOS, while planning broader integration across its vehicle models.

Hey, big spender – OpenAI has a new SKU just for you Aug. 10, 2026, 9:25 p.m.

The Register AI/ML

OpenAI has introduced a new subscription tier priced at $125 per month. The launch comes amid ongoing discussions regarding the growing capabilities of competing open-weight models.

What building an AI-native finance function taught me Aug. 10, 2026, 5 p.m.

OpenAI News

OpenAI CFO Sarah Friar has outlined five key lessons learned from building an AI-native finance organization. Her insights span automated forecasting, enhanced internal controls, and measuring return on investment for AI implementations.

Model ML completes finance work more efficiently with GPT-5.6 Sol Aug. 10, 2026, noon

OpenAI News

Model ML is utilizing OpenAI's GPT-5.6 Sol model to streamline financial workflows. The system assists across the entire process, from initial research and analysis to generating editable, traceable PowerPoint presentations and Excel workbooks.

Expanding Daybreak as the Cyber Defense Window Narrows Aug. 10, 2026, 10 a.m.

OpenAI News

OpenAI has introduced GPT-5.6-Cyber, a specialized model designed for cybersecurity tasks. It is accessible through Daybreak Red to conduct authorized vulnerability research, security testing, and exploit validation.

Putting frontier cyber models in more trusted hands Aug. 10, 2026, 10 a.m.

OpenAI News

OpenAI is granting approved Daybreak partners access to its frontier cyber models. These partners can utilize the models to provide authorized, governed cybersecurity services directly to their customers.

Tencent reportedly makes WorkBuddy a top strategic AI priority Aug. 10, 2026, 7:23 a.m.

TechNode (China)

Tencent has reportedly made its desktop AI office agent, WorkBuddy, a top strategic priority by increasing advertising, computing, and organizational resources. The promotional effort includes major mobile app placements and outdoor ads in Beijing and Shenzhen. The tool assists users by breaking down complex instructions into multi-step workflows.

Former ByteDance robotics head reportedly joins Xiaomi Aug. 10, 2026, 7:22 a.m.

TechNode (China)

Former ByteDance robotics lead Kong Tao has reportedly joined Xiaomi to head a team building foundation models for robots. Sources report that Kong made the move in 2025 alongside several former ByteDance colleagues, joining a robotics division that employs around 200 people.

WeChat tests AI-assisted writing and comments in Moments Aug. 10, 2026, 7:07 a.m.

TechNode (China)

WeChat is conducting a limited test of new AI-assisted writing and automated comment features within Moments. Integrated with the Xiaowei AI assistant, the update adds an "AI Writing" button to the publishing interface to help users generate post copy.

NEC tests parking tech that only starts charging once you exit your car Aug. 10, 2026, 1:32 a.m.

The Register AI/ML

NEC is testing new parking technology designed to begin charging fees only after a driver exits their vehicle. The report also highlights other regional developments, including Infosys signing a deal with Crocs and a 2GW data center launch in western China.

Premium seats are coming to ChatGPT Business Aug. 10, 2026, midnight

OpenAI News

OpenAI has introduced Premium seats for ChatGPT Business subscribers. The tier provides teams with five times more usage, removes five-hour usage limits, and offers flexible seating options for employees.

How Zapier transformed core marketing processes with ChatGPT Work Aug. 10, 2026, midnight

OpenAI News

Zapier's enterprise marketing team has incorporated ChatGPT Work into its core operational workflows. The tool is used to generate campaign assets, automate reporting, and decrease lead drop-offs throughout the marketing funnel.

Virgin Atlantic sharpens customer journeys with ChatGPT Work Aug. 10, 2026, midnight

OpenAI News

Virgin Atlantic is deploying ChatGPT Work across its teams to improve decision-making, product planning, and research. The tool helps the airline connect signals across the customer journey to enhance service.

How HSP GRUPPE builds AI capabilities for tax advisory Aug. 7, 2026, 9 a.m.

OpenAI News

HSP GRUPPE is integrating ChatGPT Enterprise into its operations to enhance tax advisory and client services. The adoption aims to boost productivity, improve work quality, and expand operational capacity.

DeepSeek Takes RMB141 Million Strategic Placement in Unitree IPO Aug. 7, 2026, 6:28 a.m.

TechNode (China)

Unitree Robotics has set the share price for its Shanghai STAR Market IPO at RMB150.80, issuing 40.45 million shares. Hangzhou DeepSeek AI Basic Technology Research secured a strategic placement of 933,900 shares valued at approximately RMB141 million, which is subject to a 36-month lockup period.

Humans in the loop miss a third of dangerous AI coding agent requests Aug. 6, 2026, 4:44 p.m.

The Register AI/ML

Research shows that human overseers fail to catch roughly a third of dangerous requests made by AI coding agents. This gap presents serious security risks, such as letting agents expose sensitive credentials or system configurations.

PokeBot Raises Hundreds of Millions of Dollars in Pre-A Funding Aug. 6, 2026, 6:15 a.m.

TechNode (China)

Chinese embodied-AI startup PokeBot has raised hundreds of millions of dollars in a pre-A funding round led by Shunwei Capital and Matrix Partners. Founded only months ago, the company is building household robots for chores such as cooking and folding clothes, recently demonstrating a robot that cooked mapo tofu in nine minutes.

StepFun Reportedly Separates Its Model and Agent-Device Businesses Aug. 6, 2026, 6:02 a.m.

TechNode (China)

Chinese AI startup StepFun is reportedly restructuring into two distinct divisions focused on foundation models and agent devices. Under the unconfirmed plan, the company's phone operations will spin off into a separate entity while the primary firm manages the model business. StepFun is also reportedly assembling an overseas commercialization team.

Nvidia Reportedly Seeks a Chinese Partner to Develop AI-Powered 6G Base Stations Aug. 6, 2026, 5:58 a.m.

TechNode (China)

Nvidia is reportedly seeking a Chinese base-station manufacturer to integrate its AI computing platform with wireless networking hardware for overseas 6G systems. Supply chain sources indicate Nvidia has been in discussions and collaborating with Shenzhen-based Jiaxian Communications since 2025.

Zhang Yiming Says ByteDance’s Seed Team Won’t Rely on AI Distillation Aug. 6, 2026, 5:38 a.m.

TechNode (China)

ByteDance founder Zhang Yiming reportedly directed the company's Seed AI research team to avoid relying on model distillation, even if the decision causes the firm to temporarily trail competitors. AI distillation involves training smaller models using the outputs of larger systems.

DeepSeek Plans Significant API Price Increases Aug. 6, 2026, 5:36 a.m.

TechNode (China)

DeepSeek has issued a notice informing users that it plans to implement a substantial price hike for its API services in the near future. The company has not yet released an updated price schedule, leaving the planned adjustments pending.

From asking to doing: How the world is putting ChatGPT to work Aug. 6, 2026, midnight

OpenAI News

OpenAI has published new Signals data examining how individuals and organizations use ChatGPT across the globe. The findings offer country-level perspectives on adoption rates, usage patterns, and shifts from asking questions to executing tasks.

Xiaomi open-sources embodied-AI foundation model Xiaomi-Robotics-1 Aug. 5, 2026, 8:30 a.m.

TechNode (China)

Xiaomi has open-sourced Xiaomi-Robotics-1, its foundation model for embodied artificial intelligence. The release includes the complete pipeline from real-robot post-training to deployment, as well as benchmark evaluation code. The model was trained on more than 100,000 hours of UMI data and over 10,000 hours of cross-embodiment data.

DeepSeek reportedly reopens talks on a RMB50 billion second funding round Aug. 5, 2026, 7:07 a.m.

TechNode (China)

DeepSeek has reportedly reopened discussions for a second funding round targeting RMB50 billion, which could value the Chinese AI company at roughly RMB500 billion pre-money. An agreement could potentially be reached in late August, though terms and timing have not been finalized. The talks remain unconfirmed and do not yet represent a completed deal.

DeepSeek V4 Flash tops OpenRouter weekly ranking with 7.22 trillion tokens Aug. 5, 2026, 6:35 a.m.

TechNode (China)

DeepSeek V4 Flash led OpenRouter's weekly model-usage rankings from July 27 to August 2 after processing 7.22 trillion tokens on the aggregation platform. Additionally, the model handled 8 trillion tokens on OpenCode on August 1 alone across free trials and paid developer usage.

TSMC’s 2nm monthly capacity to reach 100,000 wafers Aug. 5, 2026, 5:55 a.m.

TechNode (China)

TSMC is adjusting its manufacturing capacity plans to meet surging demand as AI clients transition to 3nm and 2nm nodes for enhanced performance and efficiency. Driven by strong order momentum from major customers, the foundry's 2nm monthly capacity is set to reach 100,000 wafers, according to supply chain sources cited by Taiwan's Economic Daily News.

Third-party cyber evaluations involving OpenAI models Aug. 4, 2026, 7 p.m.

OpenAI News

OpenAI has addressed recent incidents involving third-party cybersecurity evaluations of its models. In response, the organization outlined new safeguards designed to strengthen AI model testing and evaluation procedures.

Embodied-AI startup PaXini raises RMB1 billion to scale tactile-sensing technology Aug. 4, 2026, 8:52 a.m.

TechNode (China)

Embodied-AI startup PaXini completed a RMB1 billion strategic funding round, raising its total cumulative financing to RMB3.5 billion. The round was jointly led by an unnamed global consumer electronics and semiconductor group alongside BOC International Investment, Kunpeng Fund, and Hexin Fangce, with support from several other investment funds.

Alibaba launches next-gen Qwen3.8 with 2.4 trillion parameters Aug. 4, 2026, 7:50 a.m.

TechNode (China)

Alibaba has released Qwen3.8, its next-generation foundation model featuring 2.4 trillion parameters, offering upgraded performance in coding and workplace tasks. The model currently ranks second behind Anthropic's Claude series in third-party Arena rankings, and its API is now accessible via the Qwen AI platform.

UK mulls making employers ask before installing bossware Aug. 4, 2026, 7:30 a.m.

The Register AI/ML

The UK is considering regulations that would require employers to obtain worker consent before deploying workplace monitoring tools. The proposals encompass AI productivity tracking, keystroke logging, biometrics, and other employee surveillance methods.

DeepSeek-V4-Flash API launches on China’s National Supercomputing Internet Aug. 4, 2026, 2 a.m.

TechNode (China)

DeepSeek-V4-Flash has launched in public beta, with immediate API access and model downloads available via China's National Supercomputing Internet. Developers can use the API through the platform's Model Services portal or download the model directly through its AI community.

New ways to learn and teach with ChatGPT Work and Codex Aug. 4, 2026, midnight

OpenAI News

New education plugins for ChatGPT Work and Codex have been introduced to assist educators and students. The tools are designed to aid K–12 teachers, college faculty, and learners in teaching, research, and technical development.

Apple is getting this wrong Aug. 3, 2026, 10 p.m.

OpenAI News

OpenAI has issued a response addressing what it terms a baseless lawsuit filed by Apple. The company shared documented messages to clarify the situation and dispute allegations concerning its employees.

Insider says report of Moonshot AI filing for Hong Kong IPO this month is untrue Aug. 3, 2026, 9:57 a.m.

TechNode (China)

A source familiar with the matter has denied reports that Moonshot AI plans to apply for a Hong Kong initial public offering as soon as this month. The rebuttal comes from the insider rather than an official statement by Moonshot AI, and there is no indication that any listing paperwork has been submitted to the Hong Kong Stock Exchange.

How we built a realtime system for responsive voice AI in six months Aug. 3, 2026, 7 a.m.

OpenAI News

OpenAI has introduced GPT-Live, a real-time system developed in six months to enable continuous voice interactions with AI. The system relies on a turnless speech model and low-latency architecture to support faster and more natural conversations.

Alibaba launches Qwen3.8 with 2.4 trillion parameters Aug. 3, 2026, 4:44 a.m.

TechNode (China)

Alibaba has launched Qwen3.8, a foundation model with 2.4 trillion parameters focused on coding and professional office tasks. The model is currently accessible through the Qwen AI platform API and integrated into the new Qwen Office agent. Alibaba also plans to open-source Qwen3.8-Max next week and introduce Qwen3.8-27B.

{# Ad rails are position:fixed (see app.css's .sponsor-rail), so moving them after
in source order is purely cosmetic for the DOM -- visually they stay pinned left/right exactly as before. Put here so a linear text reader (an automated site-verification crawler, a screen reader, a search engine) hits the real page content first instead of a wall of repeated "Open slot" ad-placeholder filler before ever reaching a sentence about what Nicheloom is. #}