We introduce Fraud-R1, a benchmark designed to evaluate LLMs' ability to defend against internet fraud and phishing in dynamic, real-world scenarios. Fraud-R1 comprises 8,564 fraud cases sourced from phishing scams, fake job postings, social media, and news, categorized into 5 major fraud types. Unlike previous benchmarks, Fraud-R1 introduces a multi-round evaluation pipeline to assess LLMs' resistance to fraud at different stages, including credibility building, urgency creation, and emotional manipulation. Furthermore, we evaluate 15 LLMs under two settings: 1. Helpful-Assistant, where the LLM provides general decision-making assistance, and 2. Role-play, where the model assumes a specific persona, widely used in real-world agent-based interactions. Our evaluation reveals the significant challenges in defending against fraud and phishing inducement, especially in role-play settings and fake job postings. Additionally, we observe a substantial performance gap between Chinese and English, underscoring the need for improved multilingual fraud detection capabilities.

Keyword extraction involves identifying the most descriptive words in a document, allowing automatic categorisation and summarisation of large quantities of diverse textual data. Relying on the insight that real-world keyword detection often requires handling of diverse content, we propose a novel supervised keyword extraction approach based on the mixture of experts (MoE) technique. MoE uses a learnable routing sub-network to direct information to specialised experts, allowing them to specialize in distinct regions of the input space. SEKE, a mixture of Specialised Experts for supervised Keyword Extraction, uses DeBERTa as the backbone model and builds on the MoE framework, where experts attend to each token, by integrating it with a recurrent neural network (RNN), to allow successful extraction even on smaller corpora, where specialisation is harder due to lack of training data. The MoE framework also provides an insight into inner workings of individual experts, enhancing the explainability of the approach. We benchmark SEKE on multiple English datasets, achieving state-of-the-art performance compared to strong supervised and unsupervised baselines. Our analysis reveals that depending on data size and type, experts specialize in distinct syntactic and semantic components, such as punctuation, stopwords, parts-of-speech, or named entities. Code is available at: https://github.com/matejMartinc/SEKE_keyword_extraction

KABUL, Afghanistan – An Afghan official says two U.S. drone strikes this week hit a building where dozens of Taliban were meeting in southeastern province of Ghazni, killing 26 insurgents and wounding 22. Arif Noori, a provincial spokesman, says the drone assault was carried out on Tuesday during an offensive in Ghazni province's Andar district where the U.S. military came to the aid of Afghan security forces battling the Taliban. He says among those killed was a senior commander, known only as Aqhani, who ran the insurgency in Ghazni province. The first U.S. drone strike hit the building, the second struck an hour later after more Taliban had gathered at the scene.

DERA ISMAIL KHAN, Pakistan – Pakistani intelligence officials say a suspected U.S. drone strike has hit a militant compound near the Afghan border, killing three militants. Two officials say the unmanned drone fired two missiles at the Ghazni compound of the militant Haqqani network's commander Abdur Rasheed early in the morning on Thursday. The network is affiliated with the Taliban. They said it's unclear if Rasheed was at the compound located on the Pesho Ghar mountain in the Kurram tribal region's Ghuzgari area. The officials spoke on condition of anonymity because they are not authorized to speak to the media.