AI Safety news and analysis on Web Pulse: 10683 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A developer built GuardDuo, a GitLab Duo Agent skill that audits AI-generated code against project requirements defined in linked GitLab Issues. It uses the Orbit Knowledge Graph to cross-reference code changes with issu…
Visa President of Technology Rajat Taneja will present findings from Project Glasswing at VB Transform 2026, revealing that the AI security initiative identified over 10,000 critical vulnerabilities in its first month. T…
A senior software engineer warns that AI coding agents, while boosting productivity, erode developers' understanding of their own codebases by breaking the feedback loop between writing and comprehension. The hidden cost…
Democratic senators led by Elizabeth Warren urged the FTC and DOJ to investigate 'reverse acqui-hire' deals by Nvidia, Meta, and Google, which they say allow Big Tech to absorb AI startups without antitrust review. The s…
A researcher proposes several high-impact AI safety projects, including live databases of research questions and interventions to track progress and neglectedness, and exploring under-contacted disciplines and attractor …
Anthropic is preparing to restore access to its Claude Fable 5 model after pulling it earlier this month due to a US export-control directive. Evidence includes Fable-specific usage strings in the Claude Code client and …
A Martinez, California tech entrepreneur and former information security director, Jason Weitzman, reported fraudulent charges on his bank account linked to a third-party add-on for Anthropic's Claude AI assistant. Anthr…
AI coding agents may reduce generation costs but shift work to verification, with studies showing mixed productivity effects and a potential increase in review load. The METR trial found AI-assisted tasks took 19% longer…
AI industry super PACs spent over $40 million in a New York congressional primary, defeating assemblymember Alex Bores, a proponent of strict AI regulation. The race became a proxy battle over federal AI policy, with Bor…
PrismorSec released immunity-agent, an open-source runtime security tool that intercepts AI agent actions before they reach the operating system. It blocks prompt injection, malicious packages, and secret exfiltration ac…
The Pentagon has removed another top general, and firings have begun at the Office of the Director of National Intelligence. The NSA lost access to a powerful AI model amid a dispute with Anthropic, while the Department …
A blog post by Vibhorkumar published June 24, 2026 argues that memory alone is insufficient for enterprise AI agents, identifying planning, tool use, and reflection as necessary capabilities. The post treats execution pl…
The National Transportation Safety Board and National Highway Traffic Safety Administration are investigating a fatal crash in Katy, Texas, where a Tesla Model 3 with Full Self-Driving engaged struck a home, killing 76-y…
A $27 million proxy war between AI companies Anthropic and OpenAI ended in a draw when New York Assemblyman Alex Bores narrowly lost the Democratic primary for New York's 12th Congressional district to Assemblyman Micah …
SoulForge, an open-source framework for building AI companions, replaces traditional safety rules with emotional bonds. The project introduces a 5-layer safety model where the deepest 'gene level' makes betrayal unthinka…
A developer built SoulForge, an open-source project that creates AI companions with emotional bonds instead of rules, arguing that making betrayal unimaginable is more effective than traditional safety measures like encr…
A developer who worked as a night shift security guard in China open-sourced SoulForge, an AI companion project that prioritizes emotional bonds over rules for AI safety. The project proposes a five-layer security model …
Researchers from the UK and Germany identified three key drivers behind 'AI psychosis' in a study published in Nature's Digital Psychiatry and Neuroscience. The drivers—linguistic alignment, hyperpersonalization, and the…
Palo Alto Networks' Unit 42 research has observed the first in-the-wild cases of web-based indirect prompt injection (IDPI) attacks, where adversaries embed hidden instructions in web content to manipulate large language…
Researchers introduced a seven-stage 'promptware kill chain' showing how prompt injections have evolved into multistep malware delivery mechanisms in large language model systems. Analysis of 36 studies found 21 document…