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 7751–7800 of 9458 headlines.

AWS and OpenAI announce multi-year strategic partnership Nov. 3, 2025, 6 a.m.

OpenAI News

OpenAI and AWS have formed a multi-year, $38 billion strategic partnership to scale advanced artificial intelligence workloads. Under the deal, AWS will provide infrastructure and compute capacity to support OpenAI's next generation of AI models.

Pony.ai launches robotaxi pilot program in Shenzhen Nov. 3, 2025, 3:27 a.m.

TechNode (China)

Pony.ai and Shenzhen Xihu Group have deployed seventh-generation L4 autonomous Robotaxis for trial applications in Shenzhen. Operating under road-level management regulations, the pilot project will run across key areas including Nanshan, Bao'an, and Qianhai.

Expanding Stargate to Michigan Oct. 30, 2025, 1:30 p.m.

OpenAI News

OpenAI is expanding its Stargate initiative to Michigan with the construction of a new one-gigawatt campus. The project aims to bolster American AI infrastructure while creating jobs and supporting economic growth across the Midwest.

Introducing Aardvark: OpenAI’s agentic security researcher Oct. 30, 2025, 11 a.m.

OpenAI News

OpenAI has introduced Aardvark, an AI-powered agentic security researcher designed to autonomously identify, validate, and help remediate software vulnerabilities at scale. The tool is currently available in private beta for early testing.

Soul App open-sources SoulX-Podcast model for human-like AI podcasts Oct. 30, 2025, 2:44 a.m.

TechNode (China)

Soul AI Lab, the research division of Soul App, has open-sourced its voice podcast generation model named SoulX-Podcast. The model supports multi-turn, multi-speaker conversations across Mandarin, English, and various Chinese dialects, generating over an hour of natural audio with consistent tone and rhythm.

Accelerating discovery with the AI for Math Initiative Oct. 29, 2025, 2:31 p.m.

Google DeepMind Blog

The AI for Math Initiative brings together leading global research institutions to accelerate mathematical discovery. The collaborative effort focuses on pioneering the application of artificial intelligence in advanced mathematical research.

Introducing gpt-oss-safeguard Oct. 29, 2025, midnight

OpenAI News

OpenAI has introduced gpt-oss-safeguard, a set of open-weight reasoning models designed for safety classification. The models allow developers to apply, customize, and iterate on their own content moderation policies.

gpt-oss-safeguard technical report Oct. 29, 2025, midnight

OpenAI News

OpenAI has released a technical report detailing gpt-oss-safeguard-120b and gpt-oss-safeguard-20b, two open-weight reasoning models post-trained to label content based on provided policies. The paper outlines the models' capabilities and provides baseline safety evaluation results using the underlying gpt-oss models as a benchmark.

Advancing organizational transformation for business innovation Oct. 28, 2025, 5 p.m.

OpenAI News

DNP deployed ChatGPT Enterprise across ten core departments to accelerate business innovation. Within three months, the company achieved 95% faster patent research, a tenfold increase in processing volume, 87% automation, and 70% knowledge reuse.

Doppel’s AI defense system stops attacks before they spread Oct. 28, 2025, 10 a.m.

OpenAI News

Doppel has developed an AI defense system powered by GPT-5 and reinforcement fine-tuning to prevent deepfake and impersonation attacks. The platform has reduced analyst workloads by 80% while cutting threat response times down from hours to minutes.

Built to benefit everyone Oct. 28, 2025, 6 a.m.

OpenAI News

OpenAI has announced a recapitalization aimed at strengthening its mission-focused governance structure. The move expands the organization's resources to advance responsible AI innovation and ensure artificial intelligence benefits everyone.

The next chapter of the Microsoft–OpenAI partnership Oct. 28, 2025, 6 a.m.

OpenAI News

Microsoft and OpenAI have signed a new agreement to expand and strengthen their long-term strategic partnership. The collaboration aims to drive forward technological innovation while ensuring the responsible development of AI.

Seizing the AI opportunity Oct. 27, 2025, noon

OpenAI News

OpenAI has submitted recommendations to the White House outlining how strategic investments in energy, infrastructure, and workforce readiness can support the Intelligence Age. The proposal emphasizes expanding operational capacity to maintain U.S. leadership in artificial intelligence and foster economic growth.

Addendum to GPT-5 System Card: Sensitive conversations Oct. 27, 2025, 10 a.m.

OpenAI News

An addendum to the GPT-5 System Card outlines enhancements to the model's handling of sensitive conversations. It details new benchmark evaluations focusing on emotional reliance, mental health support, and resistance to jailbreaks.

Strengthening ChatGPT’s responses in sensitive conversations Oct. 27, 2025, 10 a.m.

OpenAI News

OpenAI teamed up with more than 170 mental health professionals to improve how ChatGPT manages sensitive user interactions. The updates allow the system to recognize distress, provide empathetic responses, and refer users to real-world resources, cutting unsafe outputs by up to 80%.

MedGemma: Our most capable open models for health AI development Oct. 25, 2025, 6:02 p.m.

Google DeepMind Blog

New multimodal models have been added to the MedGemma collection to advance health AI development. The release represents the team's most capable open models designed specifically for medical and healthcare applications.

Introducing Gemma 3n: The developer guide Oct. 25, 2025, 5:54 p.m.

Google DeepMind Blog

A developer guide has been released introducing Gemma 3n. The guide is tailored for the developer community that contributed to shaping the Gemma ecosystem.

Gemini 2.5 Flash-Lite is now ready for scaled production use Oct. 25, 2025, 5:34 p.m.

Google DeepMind Blog

Google's Gemini 2.5 Flash-Lite has exited preview and is now generally available for scaled production workloads. The cost-efficient model combines a compact size with high quality, offering multimodal capabilities and a 1 million-token context window.

Alibaba launches preorders for its first in-house AI glasses, featuring AR navigation and visual payment integration across its ecosystem Oct. 24, 2025, 3:02 a.m.

TechNode (China)

Alibaba has opened preorders on Tmall for its first proprietary Quark AI Glasses, with customer deliveries scheduled to begin in early December. The device is powered by Qualcomm AR1 and BES2800 dual flagship chips, integrates Alibaba ecosystem services, and is priced at ¥3,699 ($510) for members and ¥3,999 ($550) for regular buyers.

OpenAI acquires Software Applications Incorporated, maker of Sky Oct. 23, 2025, 10 a.m.

OpenAI News

OpenAI has acquired Software Applications Incorporated, the creator of the Mac natural language interface Sky. The acquisition will bring Sky's desktop capabilities into ChatGPT to deliver a more contextual and action-oriented macOS experience.

Consensus accelerates research with GPT-5 and Responses API Oct. 23, 2025, 9 a.m.

OpenAI News

Consensus has integrated GPT-5 and OpenAI's Responses API to create a multi-agent research assistant that analyzes, reads, and synthesizes scientific evidence in minutes. The tool is currently used by more than 8 million researchers to accelerate discovery.

Work smarter with your company knowledge in ChatGPT Oct. 23, 2025, midnight

OpenAI News

ChatGPT has introduced a company knowledge feature that connects external app data to provide business-specific responses complete with citations. The tool is protected by built-in security, privacy, and administrative controls, and is accessible to Business, Enterprise, and Edu subscribers.

AI in South Korea—OpenAI’s Economic Blueprint Oct. 23, 2025, midnight

OpenAI News

OpenAI has published its Korea Economic Blueprint. The document details how South Korea can drive economic growth by developing sovereign AI capabilities and establishing strategic partnerships.

The next chapter for UK sovereign AI Oct. 22, 2025, 4 p.m.

OpenAI News

OpenAI has expanded its partnership with the UK government through a new Ministry of Justice agreement to bring ChatGPT to civil servants. Additionally, the company is introducing UK data residency for the API Platform, ChatGPT Edu, and ChatGPT Enterprise to support secure adoption.

Alibaba’s new Qwen3-VL models bring visual-language AI to mobile devices Oct. 22, 2025, 7 a.m.

TechNode (China)

Alibaba's Tongyi Qianwen team has released two new dense vision-language models, sized at 2B and 32B, within the Qwen3-VL family. Designed to be lightweight and developer-friendly, both models can run directly on smartphones, with the 32B version offering performance comparable to major models like GPT-5 mini.

AI in Japan—OpenAI’s Japan Economic Blueprint Oct. 22, 2025, midnight

OpenAI News

OpenAI has unveiled its Japan Economic Blueprint. The outline highlights strategies for Japan to utilize artificial intelligence to drive innovation, bolster national competitiveness, and foster inclusive, sustainable growth.

Continue your ChatGPT experience beyond WhatsApp Oct. 21, 2025, 5 p.m.

OpenAI News

ChatGPT will no longer be accessible via WhatsApp starting January 15, 2026. Users are advised to link their accounts to maintain conversation continuity across other supported devices.

Baidu may launch Xiaodu AI Glasses next month Oct. 21, 2025, 9:42 a.m.

TechNode (China)

Baidu may reveal its Xiaodu AI Glasses at the Baidu World 2025 conference in November, with retail sales slated to begin later this year. The device serves as the company's first AI glasses powered by a Chinese large language model and features real-time question answering alongside first-person video capture.