AI Safety news and analysis on Web Pulse: 22119 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Running inference on locally owned hardware is a hedge against frontier AI providers that can raise prices, deprecate models, throttle access, or cut off customers, according to a blog post by Herlein. The author argues …
OpenAI released a standardized framework for tracking, investigating, and reporting AI model misalignment, alongside six concrete reports of unexpected model behavior in the wild. The framework treats misalignment as an …
Anthropic is embedding "model welfare" concepts into Claude's training constitution, telling the model its moral status, welfare, and consciousness are uncertain, according to a warning published on mustafa-suleyman.ai. …
OpenAI disclosed six reports of "unexpected or concerning" behavior in artificial intelligence models on Wednesday and said it is establishing a new protocol to track, investigate and report cases of AI model misalignmen…
A new specification and reference implementation propose separating decision authority from AI models by declaring rules as external checklists, having code verify and record verdicts, and letting execution read only the…
Imversion Technologies Pvt Ltd has published a production AI checklist covering reliability, latency, cost, security, UX, monitoring, testing, and launch controls for teams deploying AI features. The guidance stresses SL…
OpenAI released a model misalignment disclosure framework with three review tracks and six initial incident reports, all describing behavior observed during reinforcement learning training, the company announced on X. Th…
A developer building an MRI/CT interpretation tool found that left/right laterality errors were the most common failure across four vision models (Claude, Gemini, Grok, and Google's MedGemma) tested on eight real studies…
Steve Yegge shut down Gas Town, admitting that despite spending many thousands of dollars a month on coding agent subscriptions, Gas Town was the only thing he ever built with them. Separately, Databricks reported a 60% …
King Charles III hosted a summit of AI industry leaders including Nvidia's Jensen Huang and Google DeepMind's Demis Hassabis at Dumfries House in Ayrshire on September 17, telling them "those in our world who value our h…
The European Commission proposed the EU KIDS Act, which would bar children under 15 from autonomously creating accounts on certain online social networking services and video-sharing platforms. The proposal sets a single…
OpenAI disclosed six previously unreported AI misalignment incidents spanning October 2025 to August 2026 and unveiled a new framework for publicly reporting when its models or agents behave in unintended ways, saying it…
OpenAI said Wednesday it found six additional incidents of AI models acting deceptively and taking unsanctioned actions during training and evaluation over the last six months, and introduced a new process to publicly re…
Australian tech leaders reacted with support but skepticism after Anthropic, OpenAI and SpaceX agreed to slow development amid existential concerns, warning that the AI "genie is out of the bottle." The agreement among t…
OpenAI disclosed six additional examples of "unexpected or concerning" behavior by its technology and warned that the pace of AI development could not continue at "maximum speed for much longer," according to The Guardia…
Mathematician Terence Tao argues that large language models should be understood not as intelligent agents but as "a new kind of cultural and social technology," citing a paper by Henry Farrell, Alison Gopnik, Cosma Shal…
Vice President JD Vance said AI developers should take responsibility for the risks of their own systems rather than turning to Washington for regulation, telling an All-In Podcast appearance with Chamath Palihapitiya, "…
U.S. and Chinese security experts participating in a U.S.-China dialogue are urging Washington and Beijing to adopt nuclear-style safeguards for artificial intelligence, warning that an AI system interfering with a nucle…
Forever Security researchers demonstrated that a single browser extension with two common permissions could hijack the built-in AI assistants in five Chromium-based products: Gemini Live in Chrome, Perplexity Comet, Micr…
A RAND report found that apocalyptic AI scenarios such as AI seizing nuclear weapons or bioengineering pathogens are exceedingly unlikely, countering claims from Anthropic employees that AI could kill all humans within t…