Retrieval-Augmented Generation (RAG) is a promising approach to mitigate hallucinations in Large Language Models (LLMs) for legal applications, but its reliability is critically dependent on the accuracy of the retrieval step. This is particularly challenging in the legal domain, where large databases of structurally similar documents often cause retrieval systems to fail. In this paper, we address this challenge by first identifying and quantifying a critical failure mode we term Document-Level Retrieval Mismatch (DRM), where the retriever selects information from entirely incorrect source documents. To mitigate DRM, we investigate a simple and computationally efficient technique which we refer to as Summary-Augmented Chunking (SAC). This method enhances each text chunk with a document-level synthetic summary, thereby injecting crucial global context that would otherwise be lost during a standard chunking process. Our experiments on a diverse set of legal information retrieval tasks show that SAC greatly reduces DRM and, consequently, also improves text-level retrieval precision and recall. Interestingly, we find that a generic summarization strategy outperforms an approach that incorporates legal expert domain knowledge to target specific legal elements. Our work provides evidence that this practical, scalable, and easily integrable technique enhances the reliability of RAG systems when applied to large-scale legal document datasets.

We introduce Graph of Thoughts (GoT): a framework that advances prompting capabilities in large language models (LLMs) beyond those offered by paradigms such as Chain-of-Thought or Tree of Thoughts (ToT). The key idea and primary advantage of GoT is the ability to model the information generated by an LLM as an arbitrary graph, where units of information ("LLM thoughts") are vertices, and edges correspond to dependencies between these vertices. This approach enables combining arbitrary LLM thoughts into synergistic outcomes, distilling the essence of whole networks of thoughts, or enhancing thoughts using feedback loops. We illustrate that GoT offers advantages over state of the art on different tasks, for example increasing the quality of sorting by 62% over ToT, while simultaneously reducing costs by >31%. We ensure that GoT is extensible with new thought transformations and thus can be used to spearhead new prompting schemes. This work brings the LLM reasoning closer to human thinking or brain mechanisms such as recurrence, both of which form complex networks.

Tesla is being criticized by motor-safety experts for using YouTubers to beta test its self-driving technology rather than trained safety drivers. After signing non-disclosure agreements, these would-be influencers film their experiences on the road, Vice first reported, using Tesla's Full Self-Driving (FSD) Beta to navigate busy streets. Urban policy expert David Zipper criticized the FSD in a tweet, posting a clip which sees one Seattle beta tester's steering wheel suddenly spin right and the car lurches toward a crosswalk. 'Whoa, s--t!, sorry--it gave up there,' the driver announces with an apologetic wave to pedestrians. Seattle YouTuber'HyperChange,' posted a video trying out v.10 of Tesla's Fully Self Driving beta.

Security researchers say a misconfigured server owned by the controversial facial recognition company, Clearview AI, exposed its software's source code as well as internal credentials and keys. According to TechCrunch, which first reported on the flaw, Mossab Hussein, the chief security officer at SpiderSilk, a security firm based in Dubai, uncovered a flawed Clearview server storing sensitive data, allowing users to bypass its password protection. Specifically, Hussein found that a misconfiguration allowed anyone to register as a new user and access the database containing Clearview's code regardless of whether they had entered password. TechCrunch reports that, in addition to source code that would allow anyone to use Clearview's software, the database also contained passwords and other keys that would allow one to access the company's cloud storage buckets. Finished versions of Clearview's apps for iOS and Android as well as pre-developer beta versions were contained in those buckets, TechCrunch reports.

So much of our day-to-day lives has been given over to technology. From how we bank to how we shop, from how we interact with friends and colleagues to how we consume media. As you would imagine the advancements just keep on coming too, and they are having a profound impact on the legal world as well as industry. Artificial intelligence is now being deployed by lawyers across the globe to support the way they provide services to clients. Nowhere else is this marriage of legal expertise and emerging tech more obvious than in the way non-disclosure agreements can now be created using Robot Lawyer LISA.

Artificial Intelligence (AI) is having a transformative effect on the business world and the $600 billion global legal services market is not immune. As AI automates basic processes, in the legal profession it promises to allow lawyers devote their time to more valuable, cost-effective, and strategic work. Consultants at McKinsey & Company estimate that 22% of a lawyer's job and 35% of a paralegal's job can be automated. However, the common perception among lawyers remains that machines cannot yet match the legal intellect of human lawyers in daily fundamentals of the profession. This assumption was tested in the first of its kind "AlphaGo"-style Study in the legal profession.

Pioneering GCs are taking control of legal spend, armed with the latest tech. Can the rest of the in-house community keep pace? If conventional law firms have been slow to embrace technology – and they have – their counterparts in-house have been barely moving. But in the last five years signs have emerged of'early adopters' in the bluechip general counsel (GC) community who are willing to do more than apply new tools at the margins. The GCs are turning to technology to reshape the way they work. Aside from the obvious efficiency benefits, the appeal to such GC pioneers is often more potent to the professional soul: control. As Reckitt Benckiser's vice president and GC Claire Debney comments: 'I want the intellectual capital in-house, in my team. I don't want to be outsourcing all of our supply agreements, distribution agreements, or our digital platform agreements. We want our people at the meetings and in the negotiations and have the external back-up if we need it.'