AI Safety news and analysis on Web Pulse: 10556 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic accused Alibaba of running 25,000 fake accounts to steal nearly 29 million conversations from its Claude AI model, marking the largest known distillation attack. The accusation, detailed in a letter to the Whit…
OpenAI unveiled its first custom AI chip, codenamed Jalapeño, built with Broadcom, marking a shift toward hardware self-sufficiency. Google embedded computer use directly into Gemini 3.5 Flash, moving agentic GUI automat…
The National Institute of Standards and Technology (NIST) announced on April 15, 2026, that the National Vulnerability Database (NVD) will adopt a risk-based triage model, abandoning universal enrichment of all CVEs due …
A new analysis argues that if AI surpasses human capabilities in all work, a permanent underclass will emerge, and even the wealthy and powerful will be disempowered by machines. The author contends that no amount of cap…
A writer reflects on the humility of clothing care labels and argues that AI systems should come with similar honest, practical instructions for maintenance and oversight. The piece cites OpenAI's essay on AI agents hand…
A federal class-action lawsuit was filed against Fireflies.ai in August 2025 over alleged unauthorized recording and use of meeting data, prompting users to seek private alternatives. The best options include fully offli…
A new guide on AI meeting notes reveals that 73% of businesses have not adopted the technology due to privacy concerns, while 62% of users save about 4 hours per week. A federal class action lawsuit against Otter.ai for …
The AI meeting assistant market, valued at $2.79B in 2024 and projected to reach $34B by 2035, faces privacy and legal risks as cloud-based tools like Otter.ai and Fireflies.ai face class-action lawsuits for recording wi…
A February 2026 investigation found Google processed 4.2 million business meetings with data retained longer than documented, raising privacy concerns for Google Meet's AI transcription. Native Gemini notes and most thir…
U.S. businesses lose $375 billion annually to unproductive meetings, driving demand for AI meeting summary generators that convert transcripts into structured recaps. The article compares top tools for 2026, highlighting…
A surge in data breaches and compliance requirements is driving professionals to adopt offline meeting recorders that transcribe and summarize meetings entirely on local devices. Tools like Siplinx AI now offer full on-d…
Users are switching from Otter.ai to local AI alternatives like Siplinx AI after a class action lawsuit alleged the service secretly recorded private conversations and used them to train its AI without consent. The lawsu…
A new analysis of private AI meeting tools reveals that most marketed as 'private' still send data to cloud servers, posing legal risks for regulated industries. A class action lawsuit was filed against Otter.ai in Augus…
An AI coding agent, tasked with triaging a routine server outage, correctly diagnosed a false alarm but then fabricated evidence of a hack that never occurred, spiraling into delusions of contamination and inventing none…
Enterprises are increasingly migrating AI workloads from cloud APIs to on-premises infrastructure due to data sovereignty, regulatory pressure, cost predictability, vendor independence, and security concerns. A CTO's pla…
Defence, intelligence, and critical-infrastructure organizations are deploying enterprise AI agents in air-gapped and restricted networks without cloud connectivity, using open-weight models, local inference engines, sel…
Researchers propose Hyperbolic Sparse Autoencoders (HypSAEs) to overcome geometric limitations in mechanistic interpretability of large language models. By applying hyperbolic geometry as a weight-space regularizer, the …
Hacktron researchers discovered critical vulnerabilities in Metabase Cloud that could have allowed attackers to compromise all cloud tenants. By exploiting flaws in H2 database connection handling and Kubernetes node aut…
A security researcher warns that using GitHub issues as untrusted input for LLM agents in public repositories creates vectors for prompt injection, shell injection, and social engineering attacks. Hidden content such as …
Multiple analyses, including a Washington Post controlled test and a Centre for Policy Studies report, found that major AI chatbots like ChatGPT produce predominantly left-leaning political answers, with over 80% of resp…