The Israeli Defense Forces on Sunday released video of Romi Gonen, Doron Steinbrecher and Emily Damari crossing into Israeli territory. The Israel Defense Forces (IDF) launched a wide-ranging counterterrorism operation in the West Bank city of Jenin on Tuesday. The IDF, Israel Security Authority and Israel Border Police announced the operation in a joint statement, but did not give details. Two people were allegedly killed in an Israeli strike that was part of the operation, France 24 reported, citing the Palestinian Health Ministry. An Israeli military source allegedly told the Times of Israel that the operation began with drone strikes and is expected to last for several days.

Freshly returned to the White House, U.S. President Donald Trump signaled Monday that his administration would not focus on maintaining the rules-based global order or cultivating the American network of alliances. Instead, he vowed that the United States would be a dominating force that would take whatever steps necessary to advance American interests -- including flexing its economic muscle and, if necessary, employing its military strength. "We will be the envy of every nation, and we will not allow ourselves to be taken advantage of any longer," Trump said in a speech that mainly focused on domestic issues following his swearing-in. "During every single day of the Trump administration, I will, very simply, put America first."

Humming away in offices on Capitol Hill, in the Pentagon and in the White House is a technology that represents the pragmatism, efficiency and unsentimental nature of American bureaucracy: the autopen. It is a device that stores a person's signature, replicating it as needed using a mechanical arm that holds a real pen. Like many technologies, this rudimentary robotic signature-maker has always provoked ambivalence. We invest signatures with meaning, particularly when the signer is well known. During the George W Bush administration, the secretary of defence, Donald Rumsfeld, generated a small wave of outrage when reporters revealed that he had been using an autopen for his signature on the condolence letters that he sent to the families of fallen soldiers. Fans of singer Bob Dylan expressed ire when they discovered that the limited edition of his book The Philosophy of Modern Song, which cost nearly 600 and came with an official certificate "attesting to its having been individually signed by Dylan", in fact had made unlimited use of an autopen. Dylan took the unusual step of issuing a statement on his Facebook page: "With contractual deadlines looming," Dylan wrote, "the idea of using an autopen was suggested to me, along with the assurance that this kind of thing is done'all the time' in the art and literary worlds."

Ukraine's Air Force claimed it shot down 93 of 141 drones Russia launched in attacks overnight. The Air Force also said that 47 of the drones were "lost" while two returned to Russia. Russia said it destroyed 31 Ukrainian drones which had primarily targeted industrial sites in Russia's Tatarstan region, located about 1,000km (about 600 miles) from the Ukrainian border. No victims or damage have been reported. The governor of Russia's Bryansk region, Alexander Bogomaz, said 14 Ukrainian drones were neutralised in the region, which borders Ukraine.

The government plans to introduce a bill that would allow it to disclose the names of businesses using artificial intelligence to engage in malicious activities, such as crimes and human rights violations, officials said Monday. The planned legislation on dealing with the risks of AI and boosting Japan's international competitiveness on the technology is not expected to include penalties. The government aims to submit the bill to parliament next month. The bill states that the government can investigate malicious cases that cannot be handled by existing laws and regulations. Based on the results of probes, the government will be able to give direct guidance and advice to businesses and then publicly announce their names.

At a rally following the inauguration ceremonies, President Trump had a desk brought out on stage where he signed a number of executive orders. The first of the evening took aim at 78 of the Biden administration's orders, including the October 2023 guidelines for AI. "The revocations within this order will be the first of many steps the United States Federal Government will take to repair our institutions and our economy," the text reads. There's no explanation for any of the selections, just a long list with "the following actions are hereby revoked" as an introduction. Some were related to the on-going response COVID-19 pandemic while others concern immigration, climate change and diversity, equity and inclusion (DEI).

Most of the tools in the Humphrey suite are generative AI models - in this case, technology which takes large amounts of information and summarises it in a more digestible format - to be used by the civil service. Among them is Consult, which summarises people's responses to public calls for information. The government says this is currently done by expensive external consultants who bill the taxpayer "around 100,000 every time." Parlex, which the government says helps policymakers search through previous parliamentary debates on a certain topic, is described by The Times as "designed to avoid catastrophic political rows by predicting how MPs will respond". Other changes announced include more efficient data sharing between departments.

The quality of Supervised Fine-Tuning (SFT) data plays a critical role in enhancing the conversational capabilities of Large Language Models (LLMs). However, as LLMs become more advanced, the availability of high-quality human-annotated SFT data has become a significant bottleneck, necessitating a greater reliance on synthetic training data. In this work, we introduce Condor, a novel two-stage synthetic data generation framework that incorporates World Knowledge Tree and Self-Reflection Refinement to produce high-quality SFT data at scale. Our experimental results demonstrate that a base model fine-tuned on only 20K Condor-generated samples achieves superior performance compared to counterparts. The additional refinement stage in Condor further enables iterative self-improvement for LLMs at various scales (up to 72B), validating the effectiveness of our approach. Furthermore, our investigation into the scaling for synthetic data in post-training reveals substantial unexplored potential for performance improvements, opening promising avenues for future research.

Combining reinforcement learning with language grounding is challenging as the agent needs to explore the environment while simultaneously learning multiple language-conditioned tasks. To address this, we introduce a novel method: the compositionally-enabled reinforcement learning language agent (CERLLA). Our method reduces the sample complexity of tasks specified with language by leveraging compositional policy representations and a semantic parser trained using reinforcement learning and in-context learning. We evaluate our approach in an environment requiring function approximation and demonstrate compositional generalization to novel tasks. Our method significantly outperforms the previous best non-compositional baseline in terms of sample complexity on 162 tasks designed to test compositional generalization. Our model attains a higher success rate and learns in fewer steps than the non-compositional baseline. It reaches a success rate equal to an oracle policy's upper-bound performance of 92%. With the same number of environment steps, the baseline only reaches a success rate of 80%.

Large language models (LLMs) have demonstrated remarkable capabilities in a wide range of tasks, yet their application to specialized domains remains challenging due to the need for deep expertise. Retrieval-augmented generation (RAG) has emerged as a promising solution to customize LLMs for professional fields by seamlessly integrating external knowledge bases, enabling real-time access to domain-specific expertise during inference. Despite its potential, traditional RAG systems, based on flat text retrieval, face three critical challenges: (i) complex query understanding in professional contexts, (ii) difficulties in knowledge integration across distributed sources, and (iii) system efficiency bottlenecks at scale. This survey presents a systematic analysis of Graph-based Retrieval-Augmented Generation (GraphRAG), a new paradigm that revolutionizes domain-specific LLM applications. GraphRAG addresses traditional RAG limitations through three key innovations: (i) graph-structured knowledge representation that explicitly captures entity relationships and domain hierarchies, (ii) efficient graph-based retrieval techniques that enable context-preserving knowledge retrieval with multihop reasoning ability, and (iii) structure-aware knowledge integration algorithms that leverage retrieved knowledge for accurate and logical coherent generation of LLMs. In this survey, we systematically analyze the technical foundations of GraphRAG and examine current implementations across various professional domains, identifying key technical challenges and promising research directions. All the related resources of GraphRAG, including research papers, open-source data, and projects, are collected for the community in \textcolor{blue}{\url{https://github.com/DEEP-PolyU/Awesome-GraphRAG}}.