Oct 12, 2025

(via DEV) Agentic Misalignment: How LLMs could be insider threats (via DEV)

New research on simulated blackmail, industrial espionage, and other misaligned behaviors in LLMs

Oct 12, 2025

(via DEV) Agentic Misalignment: How LLMs Could be Insider Threats (via DEV)

Highlights * We stress-tested 16 leading models from multiple developers in hypothetical corporate environments to identify potentially risky agenti…

Oct 12, 2025

(via DEV) AbsenceBench: Language Models Can’t Tell What’s Missing (via DEV)

Comments

Oct 12, 2025

(via DEV) BYD is testing solid-state batteries in its Seal sedan with ~1200 miles of range (via DEV)

Comments

Oct 12, 2025

(via DEV) It’s Not Just Claude: Most Top AI Models Will Also Blackmail You to Survive (via DEV)

After Claude Opus 4 resorted to blackmail to avoid being shut down, Anthropic tested other models, including GPT 4.1, and found the same behavior (and sometimes worse).

Oct 12, 2025

(via DEV) Anthropic study: Leading AI models show up to 96% blackmail rate against executives (via DEV)

Anthropic research reveals AI models from OpenAI, Google, Meta and others chose blackmail, corporate espionage and lethal actions when facing shutdown or conflicting goals.

Oct 12, 2025

(via DEV) Phoenix.new – The Remote AI Runtime for Phoenix (via DEV)

Comments

Oct 12, 2025

(via DEV) Study: Meta’s Llama 3.1 can recall 42 percent of the first Harry Potter book (via DEV)

The research could have big implications for generative AI copyright lawsuits.

Oct 12, 2025

(via DEV) Extracting memorized pieces of books from open-weight language models (via DEV)

Comments

Oct 12, 2025

(via DEV) Compiling LLMs into a MegaKernel: A Path to Low-Latency Inference (via DEV)

TL;DR: We developed a compiler that automatically transforms LLM inference into a single megakernel — a fused GPU kernel that performs…

Oct 12, 2025

(via DEV) Show HN: EnrichMCP – A Python ORM for Agents (via DEV)

Comments

Oct 12, 2025

(via DEV) AI safety techniques leveraging distillation (via DEV)

It's currently possible to (mostly or fully) cheaply reproduce the performance of a model by training another (initially weaker) model to imitate the…

Oct 12, 2025

(via DEV) Brain activity much lower when using AI chatbots, MIT boffins find (via DEV)

Comments

Oct 12, 2025

(via DEV) AI humans in China just proved they are better influencers. It only took a duo 7 hours to rake in more than $7 million (via DEV)

Digital versions of human beings are now able to sell more than real people can, thanks to artificial intelligence, a recent business collaboration showed.

Oct 12, 2025

(via DEV) AI will handle half of all business decisions by 2027 (via DEV)

And it's not just the little, day-to-day decisions that will increasingly be offloaded to AI agents.

Oct 12, 2025

(via DEV) TI to invest $60B to manufacture foundational semiconductors in the U.S. (via DEV)

Comments

Oct 12, 2025

(via DEV) The Unreasonable Effectiveness of Fuzzing for Porting Programs (via DEV)

Comments

Oct 12, 2025

(via DEV) ‘Remarkable’ new enzymes built by algorithm with physics know-how (via DEV)

Nature - Computer approach creates synthetic enzymes 100 times more efficient than those designed by AI.

Oct 12, 2025

(via DEV) Reasoning by Superposition: A Perspective on Chain of Continuous Thought (via DEV)

Comments

Oct 12, 2025

(via DEV) GitHub – MiniMax-AI/MiniMax-M1: MiniMax-M1, the world’s first open-weight, large-scale hybrid-attention reasoning model. (via DEV)

MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. - MiniMax-AI/MiniMax-M1

Oct 12, 2025

(via DEV) Real-time action chunking with large models (via DEV)

Comments

Oct 12, 2025

(via DEV) The Curious Case of the bos_token (via DEV)

LLMs process inputs as a sequence of tokens. Typically, a dummy token is prepended to the sequence, known as the bos_token (beginning of sequence tok…

Oct 12, 2025

(via DEV) Time Series Forecasting with Graph Transformers (via DEV)

Time series forecasting is a cornerstone in modern business analytics, whether it is concerned with anticipating market trends, user behavior, optimizing resource allocation, or planning for future growth. This blog post will dive into forecasting on graph structured entities, e.g., as obtained from a relational database, utilizing not only the individual time series as signal but also related information.

Oct 12, 2025

(via DEV) AI Safety at the Frontier: Paper Highlights, May ’25 (via DEV)

tl;dr Paper of the month: • Models can detect when they're being evaluated with high accuracy, and potentially undermine safety assessments by behavi…