Stay ahead of the curve with insights and trends in technology and innovation.
Explore our complete collection of articles and insights.
This research introduces Diffusion-Driven Prompt Tuning (DDPT), a novel technique that utilizes a diffusion model to generate high-quality, parameter-efficient prompt embeddings for large language models, significantly outperforming existing methods in natural language to code generation by directly optimizing the LLM’s code-generation loss.
Ali BabarWe present LLMSecConfig, a novel framework that combines static analysis tools, retrieval‑augmented generation with large language models, and rigorous validation to automatically repair security misconfigurations in Kubernetes manifests.
Ali BabarWe conducted a systematic study that validated a set of 41 anti‑phishing guidelines derived from a multi‑vocal literature review. The guidelines were designed to be actionable across a range of practitioner roles—from system designers and security staff to C‑suite executives—and to cover the full spectrum of intervention stages, including training, tooling, and incident response.
Ali BabarIn this post we explore our recent study on using ChatGPT as a conversational assistant for software security tasks. We combine a perception survey from Twitter with a hands‑on vulnerability detection experiment, revealing that while developers love the speed of GPT‑style tools, the model’s outputs are often generic and only 61 % accurate. We explain the research design, highlight key results, and discuss what this means for the future of domain‑specific security LLMs.
Ali BabarExplore how AI can transform your business operations and drive efficiency.
Ali BabarShowing 6 of 18 articles