Generative AI news and analysis on Web Pulse: 15677 curated articles tracking the latest Generative AI developments, tools, and research, updated continuously from vetted sources.
Researchers propose SVEET, a framework that requires merely training on a pretrained bidirectional video diffusion model but supports high-quality streaming video editing in an auto-regressive fashion. The work systemati…
OpenAI and Anthropic have urged Australia to relax its ban on using the country's creative content to train AI models, calling in submissions to a parliamentary inquiry for a narrow conditional-approval exemption that wo…
A developer built SupportAI, a full-duplex, low-latency AI voice agent for Shopify and D2C e-commerce stores that handles real-time customer conversations over voice calls. The agent targets sub-800ms response windows, b…
Alibaba is pushing into full-modal AI with its Qwen4 and advanced video models, positioning the platform as a lower-cost alternative to Western providers for enterprise cloud customers in China and abroad, according to A…
A new method called Mira-Scene addresses the challenge of pixel-aligned layouts for generative 3D scenes, targeting the difficulty of accurately placing high-fidelity single-image 3D objects into a coherent scene. The ap…
Tripo AI launched Tripo P2.0 on September 21, a 3D generation model that natively produces quad-dominant meshes with separated parts and edge flow, supporting up to 25,000 faces for quad topology and 50,000 for triangle …
Maneshwar, a developer building the LiveReview AI code review tool, has demonstrated a retrieval-augmented generation pipeline in Go that uses Gemini's File Search instead of a vector database, requiring only two model c…
A developer reports that writing detailed Markdown specification files and having an LLM generate code from them produces reliable output, using the Qwen 3.8 27B model (Bonsai Ternary 2) for functions such as parse_setti…
A blogger documented a step-by-step workflow for generating AI videos that place real people into existing musical performances, after an AI-generated video of Sam Altman and Dario Amodei went viral on X. The author repo…
A developer published a 14-day Micro-SaaS launch blueprint that recommends a standardized stack of Next.js 16, React 19, Supabase, Stripe, and Vercel to compress a typical three-month build cycle into a two-week sprint. …
StepFun announced Step 5 Preview, a 600B total / 27B active mixture-of-experts flagship model for agentic work with 1M context and vision, and said open weights will be released on Oct 15. StepFun claims the model delive…
Engineer pj-corridor developed a generative AI-integrated personality assessment app and detailed the trade-offs involved in prompt tuning, abuse mitigation, and cost control. The developer kept rule-based scoring determ…
The San Sebastian festival's Zinemaldia Startup Challenge selected 10 finalists from 68 submissions, its largest field since launching in 2019, to pitch on Sept. 24 during Zinemaldia & Technology. The lineup spans AI, vi…
A developer tutorial published as the first of three articles explains generative AI end to end by building a 25-line language model, measuring prompt costs, and running experiments on an ordinary laptop CPU with small o…
Researchers introduced VideoGen-Agent, a reinforcement-learning approach for video generation agents, according to a new paper. The work targets video generative models' failures on prompts requiring specialized knowledg…
Researchers proposed AdaMem, a relevance-guided soft-compression framework for retrieval-augmented generation that maps learned passage-relevance estimates to a query-dependent allocation of a fixed memory-token budget, …
Researchers posted arXiv paper 2609.22107v1 proposing a generalized multimodal foundation model that handles arbitrary modality combinations and prediction tasks without task-specific adaptation. The model is trained on …
Researchers built the first benchmark for camera motion analysis, comprising a motion dataset with 11 motion styles and evaluation protocols, and proposed a motion representation that augments optical flow with vorticity…
A single-subject EEG-to-image baseline on the THINGS-EEG2 dataset achieved image recall of 12.83 +/- 0.58%, 39.17 +/- 1.76%, and 58.00 +/- 1.73% at ranks 1, 5, and 10 across three training seeds, against analytical chanc…
A unified analytical framework for streaming video diffusion shows that fully denoised history is not required for high-quality generation, according to an arXiv paper (2609.22283v1). On the full VBench prompt set, the p…