AI Safety news and analysis on Web Pulse: 10170 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Researchers from Microsoft Research, UC Berkeley, UC San Francisco, and Columbia University developed generative causal testing (GCT), a method that distills black-box brain-prediction models into testable verbal explana…
In November 2025, a frontier AI developer disclosed the first large-scale cyberattack executed largely by an AI model, manipulated by a state-linked actor, against roughly thirty organizations at machine speed. The Five …
Frontiers research integrity manager Simone Ragavooloo argues that the research community must stop conflating all AI-related issues with misconduct, as a new whitepaper reveals inconsistent definitions and polarized vie…
The United States and India are engaged in sensitive national security discussions over the rollout of Anthropic's Fable AI model, aiming for a gradual, measured approach to ensure safety for both nations and trusted par…
A new open-source repository, the Agent Engineering Roadmap, provides a structured, beginner-friendly guide for building production-ready AI agents, covering topics from single agents to multi-agent colonies and producti…
Billionaire Mark Cuban urged AI companies to spend billions of dollars to support towns and cities affected by job losses from artificial intelligence, calling it a 'cost of doing business.' In a post on X, Cuban said AI…
Snowflake introduced Horizon Context, a governed meaning layer designed to resolve semantic ambiguity in enterprise AI systems. The product addresses context fragmentation where the same term like "revenue" has different…
The Trump administration plans to restrict OpenAI's launch of its latest model, GPT 5.6, requiring government approval for the twenty partners set to receive early access. The move aims to mitigate risks of the powerful …
GitHub Copilot now supports bring-your-own-key (BYOK), allowing users to connect any AI model provider or run inference locally. The feature, announced June 23, addresses data governance concerns and provides resilience …
Researchers at arXiv find that verifying coding agent outputs is now harder than generating them, as foundation models improve. They argue that no fixed reward function remains effective as policy capability grows, and v…
A study comparing LLM-generated research methods against 1,000 arXiv computer-science papers found that LLMs suggest a much narrower range of methods, with provider choice showing the strongest imbalance. Academic models…
A new study from Japan's AI Security Institute finds that LLM-as-judge safety evaluations are not reproducible even at temperature 0, with per-item disagreement up to 50% across runs. The researchers tested 690 API calls…
Researchers introduced Know2Guess, a contamination-aware multi-zone benchmark with 1,200 items across five domains to evaluate large language models' ability to distinguish answerable knowledge from abstention-expected u…
A new study on Llama 3.1 8B finds that helpfulness post-training (SFT and GRPO) significantly degrades animal compassion values compared to coding-domain post-training, with a 35.7% vs. 65.2% gap on the Animal Harm Bench…
Researchers released ProfileFoundry, a synthetic dataset of 100,000 person objects with 709,228 events, 40,338 households, and 52,491 employers, designed for evaluating LLM agents on memory, privacy, and tool-use tasks. …
Researchers found that applying Nonviolent Communication constraints to large language models reduces conversational escalation in conflict-prone dialogues, improving trustworthiness without requiring extensive retrainin…
Researchers propose VIGIL, a reinforcement-learning post-training framework that reduces hallucinations in multimodal large language models by maximizing mutual information between visual input and generated responses. T…
Researchers propose ForeAgent, an agentic forensics framework for AI-generated image detection that uses a Perception-Verdict architecture and a Hindsight-Driven Self-Refining strategy to iteratively improve. The system …
Researchers propose a lattice-theoretic method for AI oracles to report unbiased, self-consistent probability sets, using the Knaster-Tarski fixed-point theorem to select a canonical credal set that avoids the self-refer…
Researchers introduced KG-TRACE, a neuro-symbolic framework that integrates the WHO mutation knowledge graph as a biological constraint on neural genomic models for antimicrobial resistance prediction. Evaluated on the C…