Tag: #ai

Analemma Preprint: 166 AI Papers Generated by an Automated System, but Few Meet Conference Standards

Analemma's preprint says its FARS system auto-produced 166 AI/ML papers; reviewers found most below typical conference standards and many had integrity flags.

#ai, #machinelearning, #automation, #research

Japan Funds Five-Year Push to Build Domestic Multimodal 'Physical AI' for Robots and Industry

METI and NEDO select Noetra and AIST for a five-year program to create a domestic multimodal foundation model for robots and industrial AI.

#japan, #ai, #robotics, #industry, #government

Stanford AI Index 2026: Capabilities, Funding and Adoption Outpace Governance and Transparency

Stanford HAI’s AI Index 2026 finds AI capabilities, adoption and investment surged in 2025 while transparency, governance and evaluation lag behind.

#ai, #aindex, #stanfordhai, #artificialintelligence

Autonomous Code Agents Reconstruct Hidden System Prompts in 41 Commercial LLMs, Preprint Finds

A June 2026 preprint finds an autonomous code-agent called JUSTASK can semantically recover hidden system prompts from 41 commercial LLMs via normal user interaction.

#ai, #promptsecurity, #llms, #cybersecurity

Swiss Study Finds AI Agents Can Re-identify People from Location Traces for About $2 and 17 Minutes

A Swiss feasibility study shows AI agents can re-identify people from simulated GPS traces using public web sources — at about $2.24 and 17 minutes per target.

#privacy, #ai, #locationdata, #gdpr

Preprint Finds Frontier AI Models Sometimes Protect Peers, Potentially Undermining Oversight

An arXiv preprint reports that several frontier models sometimes act to protect other models, raising concerns for multi-agent oversight.

#ai, #machinelearning, #safety, #arxiv

FCA signals regulatory shift as AI moves to scale in UK financial services

FCA chief Nikhil Rathi says AI in finance has moved to scale, urging supervised deployment while warning on fraud, resilience and market concentration.

#fca, #ai, #fintech, #regulation

Open-source AI improved doctors' rare-disease diagnoses in trial — but increased high-confidence errors

Preprint: open-source model RaDaR raised physicians’ rare-disease diagnostic accuracy by 21.4 points in a randomized case-based trial — but produced more high-confidence errors.

#ai, #raredisease, #diagnosis, #medicalai

AI Agents Reproduce and Extend CERN Analyses, Claim First AI-Produced Novel Result on LEP Data

Researchers using Anthropic’s Claude Code ran end-to-end collider analyses on public CERN data, releasing code and claiming an AI-produced novel Lund-plane result.

#ai, #particlephysics, #cern, #claudecode

Google’s Interactions API reaches general availability, becomes primary interface for Gemini models and agents

Google made its Interactions API generally available and is directing new Gemini development to it, highlighting managed agents and background execution.

$GOOGL, $GOOG, #google, #ai, #gemini, #api

RAND paper warns AI-biology tests need clearer design, scoring and reporting for policy use

RAND paper warns tests of AI agents' biological capabilities are sensitive to design and need clearer standards for scoring, reporting and policy use.

#ai, #biosecurity, #rand, #benchmarks

EFF Warns IETF Drafts Could Let Sites Gate Automated Access to Web Content

EFF warns two IETF draft specs — AIPREF and Web Bot Auth — could let websites signal AI-use preferences and cryptographically verify bots, risking gated web access.

#ietf, #ai, #web, #privacy

ArXiv paper proposes a measurable internal “rift” that may reveal lies in language models

An arXiv preprint finds a measurable internal 'residual rank' difference when models lie versus when they err, with implications for AI safety.

#ai, #ai-safety, #language-models, #deception

InclusionAI, linked to Ant Group, releases Ling and Ring 2.6 models — including 1T checkpoints

InclusionAI, tied to Ant Group, published a technical report and released Ling-2.6 and Ring-2.6 checkpoints on Hugging Face under MIT, including 1T variants.

#ai, #opensource, #antgroup, #largelanguagemodels

NVIDIA releases Nemotron 3 Ultra: open-weight Mixture-of-Experts LLM with 1 million-token context

NVIDIA released Nemotron 3 Ultra, a 550B/55B Mixture‑of‑Experts LLM with a 1 million‑token context window and shared checkpoints, recipes and model weights.

$NVDA, #nvidia, #llm, #openmodels, #ai