The Cloudflare CEO joined to talk about standing up to content scraping, the internet's potential futures, and his company's relationship to Trump. Matthew Prince may not be a household name, but the world most certainly knows his work. Prince is the cofounder and CEO of Cloudflare . Launched in 2010, the internet infrastructure company has found itself increasingly in the position of serving as the web's bodyguard. It filters out bad traffic, keeps sites safe, and stops them from crashing when too many people visit. Its tools defend against DDoS attacks. In 2017, Cloudflare made headlines when it dropped white supremacist site The Daily Stormer . Cloudflare's severing of ties with The Daily Stormer marked a momentous shift, one that came after years of claiming a neutral stance. Prince continues to evolve the way Cloudflare works. In July, the company rolled out a new tool tasked with blocking unauthorized AI scraping. It effectively creates a pay-per-crawl model requiring AI platforms to shell out money if they want access to a site's content. On this episode of, I talked to Prince about publishing, the old internet, and how his ideal version of the future web means that OpenAI just might become the Netflix of content. KATIE DRUMMOND: Good to have you here, Matthew. You should have been warned ahead of time, but you probably weren't.

Five members of the Japan Foundation's Nihongo Partners program (front) gather at the Japanese Embassy in New Delhi on Monday. NEW DELHI - Five people dispatched from Japan to assist in Japanese language education in India gathered in New Delhi on Monday for a six-month program aimed at enhancing cultural exchanges between the two countries. Under the Nihongo Partners program run by the Japan Foundation, the five will assist Japanese language teachers and introduce Japanese culture at secondary schools in the Delhi area over six months. It is the first time that Nihongo Partners are dispatched to a South Asian country, as the program has previously focused on Southeast Asia. The Japan Foundation plans to carry out a similar dispatch to India continuously over a decade starting this year, as part of an agreement reached at a summit of Japanese and Indian leaders last month to increase personnel exchanges between the two countries.

Artificial intelligence avatar Una will appear at the U.N. pavilion at Osaka Expo later this month as part of initiatives to combat climate change. An artificial intelligence avatar will appear at the U.N. Pavilion of the 2025 World Expo in Osaka in late September, sharing stories from Pacific island nations under threat from rising sea levels caused by climate change. The anime-inspired female character, Una, developed as part of climate initiatives supported by the Japanese government, will be showcased with the use of 3D hologram technology from Sept. 29 to Oct. 4. Launched online in May, Una can automatically respond to questions in English, Japanese and other languages. She will be like a strong voice to raise awareness on environment and climate, what is happening in the Pacific, Kanni Wignaraja, U.N. assistant secretary-general and regional director for Asia and the Pacific at the U.N. Development Programme (UNDP), said in an interview in Tokyo. In a time of both misinformation and too much information, quality journalism is more crucial than ever.

Is Chicago the violent crime capital of the US? How did India-US relations decline so fast? A Ukrainian drone attack killed two women in the village of Golovchino in Russia's Belgorod region, Russia's state TASS news agency reports. A man who was seriously injured in a Ukrainian drone attack in Russia's Belgorod region in April has died in hospital, TASS reports. TASS also reported that Russian forces shot down 82 Ukrainian drones in a 24-hour period.

This notebook reports the XplaiNLP submission to the CheckThat! 2025 shared task on multilingual subjectivity detection. We evaluate two approaches: (1) supervised fine-tuning of transformer encoders, EuroBERT, XLM-RoBERTa, and German-BERT, on monolingual and machine-translated training data; and (2) zero-shot prompting using two LLMs: o3-mini for Annotation (rule-based labelling) and gpt-4.1-mini for DoubleDown (contrastive rewriting) and Perspective (comparative reasoning). The Annotation Approach achieves 1st place in the Italian monolingual subtask with an F_1 score of 0.8104, outperforming the baseline of 0.6941. In the Romanian zero-shot setting, the fine-tuned XLM-RoBERTa model obtains an F_1 score of 0.7917, ranking 3rd and exceeding the baseline of 0.6461. The same model also performs reliably in the multilingual task and improves over the baseline in Greek. For German, a German-BERT model fine-tuned on translated training data from typologically related languages yields competitive performance over the baseline. In contrast, performance in the Ukrainian and Polish zero-shot settings falls slightly below the respective baselines, reflecting the challenge of generalization in low-resource cross-lingual scenarios.

We show that coherent, long-form musical composition can emerge from a decentralized swarm of identical, frozen foundation models that coordinate via stigmergic, peer-to-peer signals, without any weight updates. We compare a centralized multi-agent system with a global critic to a fully decentralized swarm in which bar-wise agents sense and deposit harmonic, rhythmic, and structural cues, adapt short-term memory, and reach consensus. Across symbolic, audio, and graph-theoretic analyses, the swarm yields superior quality while delivering greater diversity and structural variety and leads across creativity metrics. The dynamics contract toward a stable configuration of complementary roles, and self-similarity networks reveal a small-world architecture with efficient long-range connectivity and specialized bridging motifs, clarifying how local novelties consolidate into global musical form. By shifting specialization from parameter updates to interaction rules, shared memory, and dynamic consensus, MusicSwarm provides a compute- and data-efficient route to long-horizon creative structure that is immediately transferable beyond music to collaborative writing, design, and scientific discovery.

Online AI platforms for creating music from text prompts (AI music), such as Suno and Udio, are now being used by hundreds of thousands of users. Some AI music is appearing in advertising, and even charting, in multiple countries. How are these platforms being used? What subjects are inspiring their users? This article answers these questions for Suno and Udio using a large collection of songs generated by users of these platforms from May to October 2024. Using a combination of state-of-the-art text embedding models, dimensionality reduction and clustering methods, we analyze the prompts, tags and lyrics, and automatically annotate and display the processed data in interactive plots. Our results reveal prominent themes in lyrics, language preference, prompting strategies, as well as peculiar attempts at steering models through the use of metatags. To promote the musicological study of the developing cultural practice of AI-generated music we share our code and resources.

Rapid developments of large language models have revolutionized many NLP tasks for English data. Unfortunately, the models and their evaluations for low-resource languages are being overlooked, especially for languages in South Asia. Although there are more than 650 languages in South Asia, many of them either have very limited computational resources or are missing from existing language models. Thus, a concrete question to be answered is: Can we assess the current stage and challenges to inform our NLP community and facilitate model developments for South Asian languages? In this survey, we have comprehensively examined current efforts and challenges of NLP models for South Asian languages by retrieving studies since 2020, with a focus on transformer-based models, such as BERT, T5, & GPT. We present advances and gaps across 3 essential aspects: data, models, & tasks, such as available data sources, fine-tuning strategies, & domain applications. Our findings highlight substantial issues, including missing data in critical domains (e.g., health), code-mixing, and lack of standardized evaluation benchmarks. Our survey aims to raise awareness within the NLP community for more targeted data curation, unify benchmarks tailored to cultural and linguistic nuances of South Asia, and encourage an equitable representation of South Asian languages. The complete list of resources is available at: https://github.com/trust-nlp/LM4SouthAsia-Survey.

Recent advances in song identification leverage deep neural networks to learn compact audio fingerprints directly from raw waveforms. While these methods perform well under controlled conditions, their accuracy drops significantly in real-world scenarios where the audio is captured via mobile devices in noisy environments. In this paper, we introduce a novel evaluation protocol designed to better reflect such real-world conditions. We generate three recordings of the same audio, each with increasing levels of noise, captured using a mobile device's microphone. Our results reveal a substantial performance drop for two state-of-the-art CNN-based models under this protocol, compared to previously reported benchmarks. Additionally, we highlight the critical role of the augmentation pipeline during training with contrastive loss. By introduction low pass and high pass filters in the augmentation pipeline we significantly increase the performance of both systems in our proposed evaluation. Furthermore, we develop a transformer-based model with a tailored projection module and demonstrate that transferring knowledge from a semantically relevant domain yields a more robust solution. The transformer architecture outperforms CNN-based models across all noise levels, and query durations. In low noise conditions it achieves 47.99% for 1-sec queries, and 97% for 10-sec queries in finding the correct song, surpassing by 14%, and by 18.5% the second-best performing model, respectively, Under heavy noise levels, we achieve a detection rate 56.5% for 15-second query duration. All experiments are conducted on public large-scale dataset of over 100K songs, with queries matched against a database of 56 million vectors.

As large language models (LLMs) are increasingly deployed in high-stakes settings, the risk of generating harmful or toxic content remains a central challenge. Post-hoc alignment methods are brittle: once unsafe patterns are learned during pretraining, they are hard to remove. In this work, we present a data-centric pretraining framework that builds safety into the model from the start. Our framework consists of four key steps: (i) Safety Filtering: building a safety classifier to classify webdata into safe and unsafe categories; (ii) Safety Rephrasing: we recontextualize unsafe webdata into safer narratives; (iii) Native Refusal: we develop RefuseWeb and Moral Education pretraining datasets that actively teach model to refuse on unsafe content and the moral reasoning behind it, and (iv) Harmfulness-Tag annotated pretraining: we flag unsafe content during pretraining using a special token, and use it to steer model away from unsafe generations at inference. Our safety-pretrained models reduce attack success rates from 38.8\% to 8.4\% on standard LLM safety benchmarks with no performance degradation on general tasks.