OpenAI News
OpenAI has developed and open-sourced Whisper, a neural network designed for English speech recognition. The model approaches human-level robustness and accuracy in transcribing speech.
The opportunity tracker for new startups.
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.
Showing 9351–9400 of 9693 headlines.
OpenAI News
OpenAI has developed and open-sourced Whisper, a neural network designed for English speech recognition. The model approaches human-level robustness and accuracy in transcribing speech.
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
OpenAI News
OpenAI has introduced an outpainting feature for DALL路E that allows users to expand generated images to any size. The tool is designed to broaden creative storytelling possibilities by extending existing artwork.
Hugging Face Blog
OpenAI News
OpenAI has detailed its approach to AI alignment, emphasizing improvements in learning from human feedback and assisting humans with AI evaluation. The overarching goal is to create an aligned AI system that can help resolve broader alignment challenges.
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
OpenAI News
OpenAI has launched an updated content moderation tool for API developers. The new Moderation endpoint improves upon the previous content filter and is available free of charge.
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
OpenAI News
Hugging Face Blog
Hugging Face Blog
OpenAI News
Hugging Face Blog
Hugging Face Blog
OpenAI News
DALL路E has entered beta testing, with plans to invite 1 million users from the waitlist in the coming weeks. Participants can create images using monthly recurring free credits or purchase additional 115-generation increments for $15.
OpenAI News
OpenAI has introduced a new technique for DALL路E 2 to mitigate bias and enhance safety. The update allows the model to generate images of people that more accurately represent the diversity of the global population.
Hugging Face Blog
OpenAI News
More than 3,000 artists across 118 countries have integrated DALL路E into their creative workflows as part of its research preview. These early access participants have uncovered novel use cases and provided feedback to guide feature development.
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
Hugging Face Blog
OpenAI News
OpenAI implemented safety mitigations during the pre-training phase of DALL路E 2 to prepare the model for wider public access. These safeguards were established to minimize risks and prevent generated images from violating content policies.
Hugging Face Blog
Hugging Face Blog
OpenAI News
Researchers trained a neural network to play Minecraft using Video PreTraining on a massive dataset of unlabeled human gameplay combined with a small amount of contractor data. After fine-tuning, the model can perform complex tasks like crafting diamond tools using standard mouse and keyboard inputs. This approach marks a step toward developing general computer-using AI agents.
Hugging Face Blog
Hugging Face Blog
OpenAI News
Hugging Face Blog
Hugging Face Blog
OpenAI News
OpenAI has developed critique-writing models to identify and describe flaws in text summaries, which significantly helps human evaluators spot errors. The research found that larger models excel at self-critiquing, with model scale benefiting critique-writing even more than summary generation. These results demonstrate the potential of using AI systems to assist humans in supervising AI performance on complex tasks.
OpenAI News
Large neural networks underpin many recent breakthroughs in artificial intelligence. However, training these models presents complex engineering and research challenges that require orchestrating GPU clusters to execute synchronized calculations.