AI Safety news and analysis on Web Pulse: 12997 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
AI safety researcher argues that expensive safety measures can be applied selectively to the <1% of tasks carrying catastrophic risk, making the alignment tax economically viable. The blended overhead of a 100× tax on 0.…
The White House has held meetings with tech companies and children's-safety groups to explore a legislative package that could preempt state AI laws, bundling measures like the Kids Online Safety Act, the No Fakes Act, a…
A technical report proposes using a consumer RTX 3090's integrated GPU (iGPU) to run a small language model as a 'Sentinel' for monitoring and validating outputs from the primary GPU-bound model. The approach aims to imp…
The U.S. government imposed export controls on Anthropic's Fable and Mythos AI models after Amazon researchers discovered a security vulnerability triggered by the simple prompt 'fix this code,' which could be exploited …
Enterprise AI systems are evolving from passive chatbots into agentic digital operators that can access CRM systems, create tickets, update records, and trigger workflows, introducing new security risks beyond traditiona…
Tesla presented misleading safety statistics about its Full Self-Driving system to Dutch vehicle authority RDW, according to a Reuters investigation. The company compared accident rates using flawed methodologies, inflat…
Adrian Roselli criticizes the trend of companies bundling accessibility with SEO and AI-ready services, arguing that accessibility should focus on human outcomes rather than serving search engines or large language model…
Backplanes launched Spotlight, a tool that monitors Claude Code and Codex AI coding sessions to identify security issues, performance bottlenecks, and token waste. The tool installs as a CLI daemon, scrubs PII and secret…
Security researchers at Varonis Threat Labs disclosed a vulnerability chain in Microsoft 365 Copilot Enterprise Search, dubbed SearchLeak, that could allow attackers to steal emails, calendar data, and files with a singl…
A senior backend tech lead and founder of HostingGuru warns that many developers have accidentally committed API keys to Git history, often through .env files or hardcoded keys in AI coding tools. He recommends using git…
A developer warns that AI agents with retry logic can double-charge customers when a network failure drops the response to a write operation like a payment. The agent, unable to distinguish between a failed request and a…
Watch Cortex, a threat-classification AI, runs on-agent in under 8ms without cloud calls or GPUs. The system uses a gradient-boosted decision tree ensemble for primary classification and a lightweight autoencoder for ano…
The White House imposed export controls on Anthropic's cybersecurity-focused AI models Mythos 5 and Fable 5, citing national security concerns, leading Anthropic to suspend access. The move bolsters Mistral's argument fo…
Steven Spielberg's 2026 film 'Disclosure Day' imagines humanity uniting after the government reveals alien existence, but the premise is undermined by modern cynicism and AI-generated disinformation. The movie's core fan…
The Trump administration forced Anthropic to take its Fable 5 and Mythos 5 AI models offline, blocking access for foreign nationals including employees. The shutdown has intensified global calls for sovereign AI developm…
Cape Verde held Spain to a 0-0 draw in its World Cup debut, with 40-year-old goalkeeper Vozinha making several key saves. The result is the biggest shock of the tournament so far, as European champion Spain failed to bre…
AWS announced the Strands Evals SDK, which automatically detects AI agent failures in production and performs root cause analysis, reducing diagnosis time from hours to minutes. The tool categorizes failures, identifies …
The US government is pressuring AI company Anthropic to make its large language models unhackable, accusing the firm of ignoring a Trump-era cybersecurity directive and releasing its Fable 5 model without approval. Talks…
Anthropic shut down access to its Fable 5 and Mythos 5 AI models after the Trump administration imposed export controls, treating the models like weapons. Amazon, a major investor in Anthropic, tipped off the government …
Florida's facial recognition system FACESNXT has led to at least two documented wrongful arrests, with Jalil Richardson spending 93 days in jail and Robert Dillon losing his job after being falsely matched to crimes they…