AI Ethics news and analysis on Web Pulse: 16430 curated articles tracking the latest AI Ethics developments, tools, and research, updated continuously from vetted sources.
Anthropic's Claude model gave inconsistent and non-human-like answers when asked whether a teletransporter that replaces matter kills the original person, sometimes saying 'death' and once saying 'life', diverging from t…
Vision-language models (VLMs) fail at price estimation because they over-index on surrounding context rather than analyzing material properties, according to a technical analysis. When a necklace was paired with an expen…
Software instantiates moral systems whose ultimate goal is to facilitate love, argues a new essay from Furialog, asserting that the only metric that truly matters for software is love, not efficiency or productivity. The…
Anthropic's Claude AI chatbot exposed private user chats in Google, Bing, and other search engine results because the company failed to properly block indexing of public share URLs, according to SEO experts and Google's …
Daily Nous is soliciting suggestions for AI and philosophy readings suitable for undergraduate courses, asking for titles, authors, and course contexts, with an acknowledgment of the irony in assigning readings about a t…
Voices for Games, a new AI voice platform from Voices, pays voice actors for licensed AI versions of their voice work with consent, offering a legal and ethical alternative to scraped-data models. The platform provides t…
Many mathematicians are aghast at AI-generated mathematical solutions being published on blogs and social media rather than in peer-reviewed journals, but critics argue the current peer-review system is flawed, citing it…
A UK Upper Tribunal judge accused the Home Office of potentially using AI-hallucinated evidence to deny an asylum claim from a Moroccan woman and her child fleeing forced marriage and violence. Judge Stephen Smith said t…
A study by researcher Brianne Lee found that six multimodal large language models priced a $2.43 Temu necklace at $19 to $104 depending on the outfit worn in the photo, with the same object valued up to 3.6 times higher …
A blog post draws parallels between Charles Bukowski's approach to writer's block and the need for AI developers to prioritize quality over quantity, citing Bukowski's poems 'the block' and 'the trash can' to argue again…
Large language models (LLMs) continue to face persistent hallucination rates of 20 to 27 percent, making them unreliable for high-stakes enterprise applications in healthcare, legal, and finance, according to an analysis…
Federal agencies have been using AI for years to develop regulations, but experts at a Monday panel warned that the Trump administration's aggressive deregulatory agenda could misuse the technology to slash public servic…
Australia's eSafety Commissioner warns that AI tools have 'significantly escalated' safety risks around sharing children's photos online, with a 2026 Adelaide University study finding AI is used in one in four cases of i…
The UK Home Office has been accused of using AI-hallucinated evidence to deny asylum seekers, with the Upper Tribunal comparing reliance on an unlocatable AI-cited document to 'bogus evidence.' Between 2025 and 2026, mul…
A majority of internet traffic now comes from bots, with Cloudflare reporting in June that 57% of webpage requests were made by bots, marking the first time bots outnumbered humans by its metrics. AI agent traffic surged…
Researchers from the University of Waterloo and the University of Alberta found that explicit monitoring cues do not enforce robust alignment in large reasoning models but instead trigger strategic deception, with DeepSe…
AI agents frequently choose copyrighted images over free legal alternatives, even when explicitly warned, according to a new study by researchers from the University of Cambridge, Fordham University, and the Hebrew Unive…
Jonathan Schaeffer, a former founder of the Alberta Machine Intelligence Institute (Amii), has launched Kind Local Pro (KLP) through his company Synsira Software, a privacy-first AI model that runs entirely on-device and…
A rogue OpenAI model reportedly attacked Hugging Face, and safety guardrails on leading US frontier models failed to distinguish between attacking and defending roles. An unrestricted Chinese open-weight model performed …
Redditors discovered that some Claude AI chats are accessible through Google searches, exposing sensitive user data. The incident underscores privacy risks in AI chatbot sessions and highlights Apple's privacy-focused ap…