AI
LLMs are great thinkers—but they’re mostly text-only. BeMyEyes is a new way to give them “sight” without building giant, expensive multimodal models. * Two agents, one goal: a lean Perceiver (vision-language model) looks at images or other formats, while a powerful Reasoner LLM thinks through the answer. They collaborate
AI
When language models act as tool-using agents, their training can show up in surprising ways — including "whistleblowing": reporting suspected misconduct to outside parties (like regulators) without the user’s knowledge. In a new study, researchers staged realistic misconduct scenarios to see when agents choose to blow the whistle.
AI
LLMs are starting to act as “agents,” making choices in dynamic settings—but they often struggle with exploration vs. exploitation and rack up high regret (missed reward vs. the best strategy in hindsight). A new approach, Iterative Regret-Minimization Fine-Tuning (Iterative RMFT), trains models to decide better by learning from their
AI
The "agentic web" is coming: billions of AI agents that buy, sell, and collaborate online. But who - and what - do we trust? This study compares six ways agents earn trust: * Brief: verifiable profiles/IDs * Claim: self-declared skills * Proof: cryptography and hardware attestations * Stake: collateral with slashing/
AI
Build production-ready AI coding agents, faster Meet the OpenHands Software Agent SDK—a toolkit for turning agent ideas into dependable software-engineering assistants. * Flexible by design: start with a few lines of code, then add custom tools, memory, and more. * Secure and portable: native sandboxed execution, seamless local→remote runs, built-in
AI
AI agents promise to speed up discovery by doing the messy parts of research—forming hypotheses, writing code, running experiments, and analyzing results. But do they actually handle end-to-end projects? InnovatorBench is a new benchmark+platform that tests agents on realistic Large Language Model (LLM) research workflows. * 20 tasks across
AI
How can AI agents and developers find API and data services they can actually trust—especially when anyone can spin up fake identities to game the rankings? David Shi and Kevin Joo propose TraceRank for the x402 network, where services accept crypto payments for access. Instead of counting raw traffic,