0 views collected around this technical thread.
The article analyses how privacy concerns, data‑security priorities and an application‑first strategy cause both Apple’s Apple Intelligence and WeChat’s AI features to lag behind hype, examining product decisions, technical constraints, and the potential future of AI agents within these ecosystems.
This article presents six practical data‑desensitization methods—including string replacement, encryption, database masking, cache‑based replacement, dynamic AOP masking, and K‑anonymity—explains their principles, shows Java implementations, compares security, performance, and reversibility, and offers concrete recommendations for protecting sensitive fields such as phone numbers and ID cards.
The article presents the abstracts, publication types, links, and research areas of seventeen Ant Group papers accepted at ICLR 2025, covering topics such as embodied robot co‑design, efficient distributed training for large language models, optimization via LLMs, character animation, interactive frame interpolation, KV‑cache management, and privacy‑preserving Transformers.
The ScaleOT framework introduces a privacy‑preserving offsite‑tuning pipeline for large language models that combines importance‑aware dynamic layer replacement with selective rank compression, enabling flexible model compression, near‑lossless fine‑tuning, and strong privacy guarantees across diverse downstream tasks.
This article presents concise English summaries of the 18 Ant Group papers accepted at AAAI 2025, covering topics such as privacy‑preserving large‑model tuning, knowledge‑graph integration, AI‑generated image detection, multi‑task learning, generative retrieval, role‑playing evaluation, and video hallucination mitigation.
This article presents a curated overview of fifteen Ant Group research papers accepted at NeurIPS 2024, covering topics such as large language models, knowledge graphs, recommendation systems, privacy-preserving inference, and multimodal learning, with abstracts, paper types, links, and key contributions highlighted.
This presentation reviews recent research on data security and utilization in the large‑model era, covering privacy‑preserving federated learning, knowledge‑transfer techniques, prototype‑based modeling, multi‑model fusion methods such as FuseGen, and introduces the federated knowledge computing platform FLAIR for both horizontal and vertical federated scenarios.
This guide explains how to design and build a privacy‑focused, view‑once image sharing system using Spring Boot, MySQL, Thymeleaf, and optional cloud storage, covering requirements analysis, architecture, environment setup, code implementation, security considerations, performance optimizations, testing, and deployment.
This paper introduces FedMix, a method that enhances vertical federated learning by mixing aligned and unaligned data, theoretically demonstrating the value of unaligned data and empirically achieving over 10% ROI improvement and significant AUC gains while keeping computational and communication overhead low.
To meet tightening privacy regulations, the team is optimizing iOS dynamic privacy checks by replacing the CPU‑intensive +[NSThread callStackSymbols] method with a low‑overhead, non‑blocking stack‑symbolization solution that maintains compliance testing speed and prevents device slowdown during large‑scale automated gray‑release scans.
The article explains how many backend services use logical deletion—simply flagging or masking user data instead of physically removing it—to appear compliant with standards like China's GB/T 35273, while highlighting risks such as continued e‑bike control, fraud detection challenges, and missed logout handling.
Amidst a growing trend that narrows performance gaps between large and small language models, researchers highlight the efficiency, adaptability, and specialized advantages of small language models (SLM), while also discussing the high costs, hallucinations, and security concerns that still challenge large‑scale LLMs.
Exposing your public IP address can invite DDoS floods, vulnerability scanning, privacy breaches, and targeted phishing or malware attacks, so you should protect it with firewalls, regular updates, secure VPN connections, and by avoiding public disclosure on social platforms.
This article describes how ZhaiZhai implements log desensitization for personal data protection using a reusable Java utility class, JSON serializer customizations with Jackson, and an annotation‑processing tool that automatically generates masked toString methods, while also discussing operational challenges and future plans.
LocalAI is an open‑source, cost‑effective alternative to the ChatGPT API that lets you download and run thousands of language models locally via Docker or compiled binaries, offering privacy, customization, and easy integration into projects through a compatible API.
In late 2023, Ant Group and academic partners reported fourteen security‑focused research papers accepted at top venues such as USENIX Security, ACM CCS, and USENIX ATC, covering privacy‑preserving computation, secure two‑party GBDT training, macOS kernel fuzzing, privacy‑preserving ML frameworks, Rust OOM handling, and more.
Researchers discovered that prompting ChatGPT with repeated words can cause the model to leak private data such as phone numbers and email addresses, highlighting a serious repeat‑prompt vulnerability that reveals substantial personally identifiable information from its training corpus.
OpenAI revealed a new data partnership initiative to collect large‑scale public and private datasets across multiple modalities, aiming to improve AI model safety and usefulness by incorporating diverse, hard‑to‑access human‑generated content while respecting privacy and intent.
Weibo has begun testing a front‑page real‑name system for high‑follower accounts and a visitor‑record function limited to premium members, sparking user backlash over privacy concerns, debates about platform monetisation, and discussions of the broader impact on social media behaviour.
This article explores multiple strategies for enabling fuzzy search on encrypted phone numbers, including in‑memory caching, database decryption functions, segment‑wise storage, and dedicated fuzzy‑search fields, comparing their trade‑offs in performance, memory usage, and data consistency.