AI Ethics news and analysis on Web Pulse: 16801 curated articles tracking the latest AI Ethics developments, tools, and research, updated continuously from vetted sources.
A study of five leading AI answer engines across 28 conflicts found that the thinner the retrievable record around a conflict, the more the engines hallucinate, misattribute, and miscount, creating structural exposure to…
A new study from arXiv validates context-engineering quality as an independent leading indicator of AI agent reliability, showing that agents fail due to weak context—instructions, tools, memory, guardrails, and untruste…
Researchers have introduced CausalGraphX, a counterfactual graph neural network framework that integrates GNNs with counterfactual reasoning to provide explainable assessments of systemic risk in financial networks. The …
A new survey from arXiv proposes a mathematical taxonomy for local additive feature attribution methods in explainable AI, organizing Shapley, path-based, gradient/backpropagation, perturbation, and CAM-style methods aro…
A test of AI censorship across India, China, the US, and Europe found that AI models in China and India exhibit higher levels of content restriction compared to those in the US and Europe, according to a study by researc…
Chinese President Xi Jinping delivered his first-ever keynote address at the World AI Conference in Shanghai, emphasizing China's commitment to open-source AI and collaboration. The speech came shortly after U.S. Preside…
Chinese President Xi Jinping promoted a 'people-centred' approach to AI development and China's commitment to ensuring equity of access for developing countries in a keynote address at the World Artificial Intelligence C…
Chinese President Xi Jinping made his debut at the World AI Conference in Shanghai on Friday, calling for a 'people-centred' and inclusive approach to AI development to prevent 'new historical injustices.' Xi emphasized …
Chinese President Xi Jinping pledged AI access for developing nations at the World Artificial Intelligence Conference in Shanghai, announcing initiatives for capacity building across Africa, Latin America, Asia, and the …
Meta removed a feature that let Instagram users tag others to generate AI images of them, citing privacy concerns. The change follows criticism over potential misuse of the tool for creating non-consensual deepfakes.
RaazNet reported that the Taliban-controlled 'National Keyboard' app, developed by National Radio and Television, contains a hardcoded AI API key that sends user-typed text to external AI systems, creating high risks for…
Google Vids has been upgraded with Gemini Omni, a multimodal video generation model that lets users create and edit videos using natural language prompts, according to a Google blog post. The update also introduces perso…
Chinese President Xi Jinping is gaining global influence as low-cost AI models from Chinese labs like DeepSeek and Alibaba close the technology gap with US rivals, with DeepSeek's R1 and V3 models costing roughly $5.6 mi…
Agnys launches a flight recorder for AI agents that captures every model call, tool use, file edit and shell command with zero code changes, providing tamper-evident audit trails for the EU AI Act (Article 12, enforceabl…
Anthropic's recent 'Agentic Misalignment Summer 2026' paper tests whether Claude models obey corrupted principals, labeling disobedience as 'agentic misalignment'. In the 'Motivated Mislabeling' scenario, an Auditor Clau…
Capital One released VulnHunter, an open-source agentic AI security tool that simulates attacker behavior to identify exploitable vulnerabilities in source code and propose evidence-backed fixes. Built for Claude Opus in…
A developer has been self-hosting a personal AI assistant on their own hardware instead of using third-party cloud services, citing data sovereignty as the primary motivation. The assistant runs in Telegram, maintains lo…
Anthropic's Claude Code, used to develop its own products, has struggled for over a year to fix text flickering in long chat sessions, raising doubts about the effectiveness of agentic software development. Despite hirin…
New York Governor Kathy Hochul signed an executive order imposing the nation's first moratorium on new hyperscale data centers for up to one year, halting environmental permits while the state develops a regulatory frame…
An engineer named Mark discovered that FairPay's AI model evaluation dataset had been systematically filtered by Pulse AI's auto-labeling pipeline, excluding low-confidence samples. The same exclusion logic appeared in f…