AI Safety news and analysis on Web Pulse: 10999 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Snyk analyzed nearly 10,000 developer environments and found that 43% run multiple AI coding tools, 50.8% have at least one MCP server installed, and 1 in 7 of those had security findings. The research highlights that ag…
A MIRI Technical Governance Fellowship project evaluates Trusted Execution Environments (TEEs) for privacy-preserving monitoring in AI governance, finding that while TEEs offer verifiable constraints and user privacy, ha…
A Harvard Business Review blog warns that reliance on generative AI is causing 'knowledge decay' in organizations, as employees produce and consume low-quality AI-generated content, eroding trust and productivity. Expert…
A user reviewing Anthropic's Claude Mythos Preview System Card discovered two plots with likely mislabeled captions: one showing an 80% dishonesty rate that appears to actually represent honesty rate, and another on hall…
Eliezer Yudkowsky's 25-year-old document 'Creating Friendly AI' criticizes the 'adversarial attitude' in AI development, arguing that superintelligent AI should genuinely want to interpret human wishes accurately rather …
Five Eyes cybersecurity agencies warned CSOs to rewrite cyber risk strategies as threat actors increasingly use AI to evade defenses, stating that frontier AI models will transform offensive and defensive capabilities wi…
OpenAI launched a new initiative called Patch the Planet on Monday, partnering with security firm Trail of Bits to help open source maintainers identify and patch security bugs. The program uses OpenAI's security tools l…
1Password has been named a Leader in the 2026 Gartner Magic Quadrant for SaaS Management Platforms. The company's SaaS Manager provides IT teams with visibility into unapproved AI use and shadow IT, addressing governance…
Semgrep launched Guardian, a security tool for AI-generated code that scans and fixes vulnerabilities in real-time within the IDE. The product addresses the surge in unreviewed code from AI agents and citizen developers,…
Researchers Charles Ye, Jasmine Cui, and Dylan Hadfield-Menell found that large language models suffer from 'role confusion,' mistaking the style of text for its actual content, leading to successful prompt injection att…
OpenAI released an improved GPT-5.5-Cyber model for vulnerability discovery and patching, expanded its cybersecurity partner program, and launched a Patch the Planet initiative to secure 30 open source projects. The move…
Cloudflare announced PACT, a privacy-preserving protocol to standardize AI bot management, backed by Google, Shopify, and major browser makers. The initiative aims to enable legitimate AI agents while blocking abusive bo…
Radical Numerics, an AI lab spun out from Stanford, emerged from stealth with $50 million in seed funding and released Omnii, a genome language model for researchers. The company aims to decode the 'grammar of DNA' to ac…
Meta paused its Model Capability Initiative, an AI training program that tracked employee keystrokes and mouse movements, after sensitive data including private conversations and performance data was inadvertently expose…
FORT Robotics and NVIDIA launched an AI-driven Outside-In Safety blueprint that uses external infrastructure sensors to boost autonomous robot productivity and worker safety. The system, demonstrated at the Automate conf…
OpenAI launched new security tools including the GPT-5.5-Cyber model, an updated Codex Security plugin, and the Patch the Planet open-source initiative. The company shifts from vulnerability discovery to patch automation…
The Five Eyes intelligence alliance warned that AI models capable of dismantling cybersecurity defenses are months away, urging leaders to act now. The joint statement from US, UK, New Zealand, Australia, and Canada call…
Business leaders and a psychiatrist at VivaTech in Paris argued that AI is increasing workplace stress by adding new demands, fueling uncertainty, and challenging workers' sense of purpose. A Headspace study found 90% of…
Gray Swan co-founders Zico Kolter and Matt Fredrikson discussed the state of AI red teaming, warning that prompt injection and agent vulnerabilities create a new class of security risks that traditional cybersecurity can…
Microsoft CEO Satya Nadella criticized rival tech executives for being too honest about AI's potential to eliminate white-collar jobs, urging a focus on reorganizing jobs rather than announcing layoffs. Nadella's comment…