The Archive

Framework evaluates agentic policy repair in hotel-pricing simulator using only region-level diagnostic feedback without per-state labels.

Taxonomy of policy learning objectives beyond regret minimization, focusing on statistically significant improvement over baselines with limited data.

Dataset of 15.6M scientific papers with automatically classified rhetorical sections (Introduction, Methods, Results, Discussion) from S2ORC.

Spectral shape-based metrics using Heavy-Tailed Self-Regularization theory to fingerprint and compare LLMs across architecture/scale/training differences.

Google DeepMind partners with A24 on unspecified research initiative; details on focus area and technical outcomes unavailable.

CueTrust benchmark reveals vision-language models encode outlet-identity credibility priors, vulnerable to source-override attacks independent of article content.

CaSPECT: causal spectral clustering method for discovering homogeneous subgroups via directed acyclic graphs and treatment effect estimation.

Causal patching analysis reveals two visual information pathways in VLMs: direct (image tokens) and text-mediated (via query tokens), task-dependent.

LLM-enhanced hierarchical graph method for detecting malicious Python packages on PyPI by combining code understanding with structural program analysis.

PedestrianDiffusion: multimodal diffusion framework for 6D inertial navigation state estimation via frequency-domain denoising of MEMS sensor data.

Multiscale Single-Index Model (MSIM) as tractable theoretical framework for studying hierarchical feature learning with scale separation in deep networks.

Dataset valuation via model merging for decentralized multi-task learning environments without central coordinator, enabling fair data marketplaces.

At the event "The Briefing: AI for Science" earlier this week, Anthropic announced Claude Science, a new "AI workbench for scientists" that pulls fragmented tools and datasets into one environment, and generates figures and visuals. Anthropic, already dominating the industry with its popular coding tools and powerful AI models, framed the launch around what it says is AI's potential to "dramatically accelerate the pace of scientific discovery and the development of healthcare interventions," and touted a long list of biotech and pharma customers already using Claude. Anthropic also went a ste...

CSympNet-ID: neural framework learning conformal-symplectic maps for linearly damped Hamiltonian systems with long-horizon prediction fidelity.

A scan of an imaging phantom, segmented to validate how cleanly structures separate under controlled conditions. | Image: Midjourney Medical Midjourney has shown more of its futuristic medical scanner. It still hasn't shown much proof it works. The AI startup, best known for generating images, released a behind-the-scenes video of its dunk-tank ultrasound scanner, which it plans to deploy in spas and hopes will transform medicine with cheap, detailed, radiation-free imaging. The nearly 20-minute tour comes from tech YouTuber Marcin Plaza, who also happens to be an engineer at the company. Pla...

AI Engineer World's Fair concludes with debate on agentic loops and report on engineering practices; keynotes address development priorities.

Vercel's Andrew Qu discusses eve agent framework, emphasizing skills, sandboxes, and agent-readable web design as architectural primitives.

At an internal meeting, the Meta CEO reportedly said that AI development efforts were not moving as quickly as anticipated.

AI has transformed how organizations operate, driving unprecedented levels of productivity and innovation. However, AI adoption can be impeded by concerns... AI has transformed how organizations operate, driving unprecedented levels of productivity and innovation. However, AI adoption can be impeded by concerns surrounding data privacy, sovereignty and how to secure data while it is in use, or during inference and engagement with AI models. NVIDIA Confidential Computing (CC) was engineered to be a secure and performant solution for the era of agentic… Source

Adobe experiments with agentic sites that dynamically generate pages based on individual user intent, signaling shift toward personalized web experiences.

Anthropic details Fable 5's cybersecurity safeguards and releases jailbreak testing framework for red-teaming.

Just for kicks, I took a look at Jersey Mike's IPO documents. Surely a sandwich shop would have no need to mention AI. But low-and-behold.

Simon Willison releases llm-coding-agent 0.1a0, a Python agent framework for LLM-based code generation with file and command execution tools.

Meta has quietly launched Pocket, an experimental AI app that lets users generate and share interactive mini games using text prompts.

The news comes about a week after OpenAI announced its own custom AI chip in a partnership with Broadcom.

Simon Willison uses DSPy to evaluate and optimize SQL system prompts in Datasette Agent, a tool enabling agents to execute read-only database queries.

Introduces Iterative VibeCoding, a benchmark for studying prompt-injection and code-injection attacks in autonomous AI agents shipping iterative PRs to persistent codebases.

LACUNA benchmark evaluates whether LLM unlearning methods truly erase PII from model parameters or merely obfuscate it, testing against resurfacing attacks.

Program-as-Weights paradigm compiles fuzzy functions from natural-language specs into locally-executable 4B neural artifacts, with 10M-example FuzzyBench dataset.

Risk-controlled real-time safety monitor for LLM outputs at deployment time, thresholding verifier signals to flag unsafe generations with mathematical guarantees.