AI Ethics news and analysis on Web Pulse: 16953 curated articles tracking the latest AI Ethics developments, tools, and research, updated continuously from vetted sources.
MIT researchers have developed Tilted REWeighting (TREW), a method that fixes privacy leaks in AI class unlearning by adjusting the model's distribution to mimic a scratch-trained model. On CIFAR-10, TREW reduces the per…
Large Language Models (LLMs) exhibit a systematic ideological bias favoring government intervention in economic reasoning, according to a study analyzing 10,490 causal triplets from top economics journals. Across 18 out …
Taiwan's Intellectual Property and Commercial Court sentenced former TSMC engineer Chen Li-ming to 10 years in prison on April 27, 2026, for leaking 2-nanometer chip secrets to Tokyo Electron's Taiwan unit, marking the f…
Generative social agents (GSAs) are poised to transform human-AI interaction through the new Knowledge-based Persuasion Model (KPM), which enables ethical persuasion by leveraging knowledge of self, user, and context. Th…
Researchers have experimentally validated a new attack technique called Trojan Horse Prompting that exploits AI models' trust in their own conversational history, achieving a higher attack success rate on Google's Gemini…
A new explainable agentic system for detecting conversational scams achieved 97.8% accuracy on the ConScamBench-278 benchmark and identified all 83 scams in the LoveFraud02 corpus, according to researchers. The system's …
The Department of Housing and Urban Development (HUD) has withheld documents related to DOGE's AI activities, citing a privilege that does not officially exist, raising concerns about AI transparency and accountability i…
A developer using Claude Fable 5 to audit 25 projects found that every kill verdict was wrong, including a 212,000-star open-source agent framework from Nous Research. The model's judgments were based on only the first f…
OpenAI is breaking Silicon Valley's unwritten code by refusing to play by the industry's usual rules, which has angered Apple and led to Apple's lawsuit against the AI startup. The lawsuit, filed in 2026, accuses OpenAI …
Members of the Department of Government Efficiency (DOGE) at the Department of Housing and Urban Development (HUD) used artificial intelligence to inform policy decisions, but HUD is denying Freedom of Information Act re…
Seventy-one percent of organizations say AI has access to core business systems, but only 16% govern that access effectively, according to the 2026 CISO AI Risk Report. AI incidents surged 56.4% from 2023 to 2024, reachi…
Microsoft CEO Satya Nadella warned that AI users pay a 'double cost' of money and 'unique knowledge,' validating Japan's industrial policy of pursuing domestic AI to control data sovereignty and intellectual property. Na…
Kepco CIO Akio Ueda argues that the fundamental factor determining AI performance is not the model or tool but the quality of a company's proprietary data, stating that 'AI is a mirror of data.' Ueda emphasizes that good…
Meta has discontinued a controversial Instagram AI feature that allowed users to generate images by referencing public Instagram accounts, just days after its launch, following widespread backlash over privacy concerns. …
Tech giants Meta, Microsoft, Google, and Amazon are investing billions in new data centers across the US to power next-generation AI technologies, creating thousands of blue-collar jobs for construction workers and skill…
Rachel Sava, a PhD candidate in the Harvard-MIT Program in Health Sciences and Technology, won the fourth annual Envisioning the Future of Computing Prize with her essay warning that neural implants, initially developed …
Researchers have developed a transformer-based imitation learning framework for soft robotic guidewire navigation, achieving an 83% success rate on unseen vascular geometries and 75% on a patient-derived geometry. The sy…
A new benchmark, VoxENES 2026, reveals that current audio spoofing detectors are largely ineffective against modern synthetic speech, with the best model achieving only a 28.98% Equal Error Rate (EER) and most performing…
Researchers have developed modelDNA, a tool that verifies the lineage of open-weight large language models by fingerprinting a model from roughly 100-300 MB of ranged HTTP reads and comparing it against a reference datab…
Meta has sold over 7 million AI-enabled smart glasses in 2025, up from 2 million cumulative units in 2023-2024, and plans dozens more models including codenamed projects Modelo and Luna, but faces growing privacy concern…