All Articles
18657 articles total
The Capitoline Wolf
A Hacker News thread titled 'The Capitoline Wolf' with minimal content provided.
Anthropic becomes first AI startup to join the Frontier carbon removal coalition
Anthropic joins the Frontier carbon removal coalition to fund environmental projects.
Anthropic got hit by export rules nobody understands
The US government used export controls to force Anthropic to block access to its Fable 5 and Mythos 5 AI models for foreign nationals.
The Slate Truck's price may have leaked, starts at $24,950
Leaked pricing for the Slate Truck suggests a starting price of $24,950.
"Dangerous" AI models are coming no matter what
Opinion piece discussing the inevitable emergence of AI models with dangerous hacking capabilities.
MODE: Modality-Decomposed Expert-Level Mixed-Precision Quantization for MoE Multimodal LLMs
Introduction of MODE, a modality-decomposed mixed-precision quantization framework to reduce GPU memory costs for MoE Multimodal LLMs.
Graph neural networks at war: integrating cybersecurity and drone intelligence in the Israeli-Iranian conflict
Research on using Graph Neural Networks (GNNs) to integrate cybersecurity intrusion detection with drone response in cyber-physical systems.
TrustErase: Auditable Instant Machine Unlearning with Passport-Embedded Representations
TrustErase is presented as a verifiable, data-free machine unlearning framework using passport-embedded representations for privacy compliance.
LineageMark: Multi-user White-box Watermarking for Contribution Tracing in Model Derivation Chains
LineageMark is a white-box watermarking framework designed to trace contributions in multi-stage AI model derivation chains.
Vibrato Expression Control for Singing Voice Conversion with Improving Independent Control
VibE-SVC2 is a singing voice conversion framework that improves independent control of pitch and timbre styles, including vibrato rate scaling.
AI chemist improves a challenging reaction in medicinal chemistry
AI is being applied to medicinal chemistry to optimize challenging chemical reactions.
Seventeen Camels and Where They Can Take You
An exploration of the 'Seventeen Camels' puzzle and its mathematical/logical implications.
The Competitive Moat That AI Can't Replicate
A discussion on the unique competitive advantages (moats) that AI cannot easily replicate in business.
Cybercriminals allegedly hacked tens of thousands of Fortinet firewalls used by major companies all over the world
Cybercriminals allegedly compromised thousands of Fortinet firewalls using previously known passwords.
Social media’s next evolution: user-controlled algorithms
Social media platforms are introducing tools that allow users to customize the algorithms powering their feeds.
Prefill/Decode-Aware Evaluation of LLM Inference on Emerging AI Accelerators
A phase-aware evaluation of LLM inference performance comparing GPUs and emerging AI accelerators like GroqRack, focusing on Prefill and Decode phases.
Models Take Notes at Prefill: KV Cache Can Be Editable and Composable
Introduces a method to make KV caches editable and composable, significantly reducing latency for time-to-first-token in LLM inference.
Timestamp-Aware Spatio-Temporal Graph Contrastive Learning for Network Intrusion Detection
A self-supervised GNN-based framework for network intrusion detection that explicitly leverages real timestamps for temporal dependency learning.
An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenarios
An evaluation of non-adversarial data leakage risks in tool-using LLM agents, highlighting safety gaps in realistic enterprise scenarios.
Probing, Fusion, and Trustworthiness: A Systematic Evaluation of Foundation Model Representations for Multimodal Cancer Analysis
Systematic evaluation of foundation model representations for multimodal cancer analysis using images and transcriptomics.