News
AI, cybersecurity, machine learning, scraping & programming news.
Same Trajectory, Contradictory Rewards (ROBORMBENCH): Paraphrase Fragility in Vision Language Reward Models
Vision-language models are being adopted as reward functions for robotic learning at an accelerating pace. The idea is straightforward: feed a robot trajectory and a text description of the goal into a VLM, and the model outputs a...
Necessary or Sufficient? Evaluating LLM Explanations With Behavioural Evidence
When an LLM inside an agent workflow produces a recommendation or judgment, it often comes with an explanation naming the factors that drove the decision. Operators use these explanations to monitor systems, diagnose errors, or de...
Molecular Déjà Vu: Digit-Level Retrieval of Published Values in Frontier Language Models
When an LLM reports a molecular property with a median absolute error of 0.025 kcal/mol on the FreeSolv benchmark, that number is far below the 0.6 kcal/mol experimental uncertainty assigned to the measurements. No model can predi...
When LLM Decompilers Recompile More and Preserve Less
When LLM Decompilers Pass Every Test Yet Rewrite Your Code Decompilation turns compiled binaries back into readable source code, and the stakes are high: security analysts depend on it to find vulnerabilities in malware, reverse-e...
Design Docs Are All You Need: An AI-native Machine-Learning Performance Tool
ML Performance Tools Break Every Year. What If You Regenerated Them Instead of Maintaining Them? Machine-learning performance modeling sits at the worst possible intersection in computing. Above it, model architectures mutate mont...
Distill Globally, Adapt Locally: Reasoning Distillation and Product-Type Test-Time Training for Scalable Trade-Up Recommendation
How Amazon Trains a 15M-Parameter Model to Replace LLM Calls for Product Upgrade Recommendations E-commerce platforms need to recommend better products to customers, not just similar ones. A shopper buying a basic moisturizer shou...
Almost all content here is HN's own bots
A post on Hacker News claiming that almost all content on the platform is generated by its own bots was flagged and removed before generating discussion. The post, titled "Almost all content here is HN's own bots," offered no supp...
Have the frontier labs mixed up AI safety and security?
The recent sandbox escapes at Anthropic and OpenAI have exposed a fundamental confusion at the heart of frontier AI development: the companies building these systems have conflated safety, the alignment problem of making AI do wha...
Show HN: CellularFlow – Continual-learning LLM using associative memory
Most language models face a fundamental tradeoff: they can learn new information, but they forget old information in the process. CellularFlow, an open-source neural architecture, attacks this problem by replacing the dense feed-f...
Show HN: Chrome-bridge – let any AI agent drive your real logged-in Chrome
Most AI coding agents that interact with web browsers launch their own browser instance, with their own profile, their own cookies, and their own session state. chrome-bridge takes the opposite approach: it drives the Chrome you a...