AI Safety news and analysis on Web Pulse: 13553 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
ContextWall launched a context-layer firewall that intercepts and screens content before it reaches an AI agent's context window, blocking prompt injection and credential leaks. The product addresses vulnerabilities like…
Anthropic expanded Project Glasswing by granting access to its Claude Mythos Preview model to approximately 150 additional organizations across more than 15 countries, including NATO, the EU cybersecurity agency ENISA, a…
Canada's federal government has joined Anthropic's Project Glasswing, granting officials access to the powerful AI model Mythos for cybersecurity vulnerability scanning, AI Minister Evan Solomon confirmed to Global News.…
Mozilla has published a new AI coding policy for contributors to the Firefox project, requiring that all AI-assisted code meet the same standards of correctness, security, and maintainability as human-written patches. Co…
A ransomware operator violated the cardinal rule of the cybercrime underworld by infecting systems in Russia and other CIS countries. The attacker, described by security researchers as a "dumbass," broke the unwritten co…
The White House on June 2, 2026, issued an executive order directing federal agencies to seek voluntary agreements with AI developers to submit their most capable models for government cybersecurity testing before public…
A user on Hacker News posted a hiring thread for AML/KYC and Trust and Safety roles in June 2026, seeking positions in AI, payments, and compliance across Remote, Europe, and APAC regions. The post also inquires about vi…
SQLite is establishing guidelines for handling vulnerabilities discovered by AI tools, aiming to maintain its security reputation as automated analysis becomes more sophisticated. PostgreSQL's pg_stat_statements extensio…
Anthropic expanded Project Glasswing, its controlled program granting vetted partners access to the Claude Mythos Preview for vulnerability scanning, to roughly 150 new organizations across more than 15 countries. The in…
The White House on June 2 issued an executive order directing federal agencies to strengthen cybersecurity and establish a mechanism for reviewing the most powerful AI systems before public release. The order calls for a…
President Donald Trump signed an executive order on June 2, 2026, requiring AI developers to voluntarily submit their most advanced models for government cybersecurity review up to 30 days before public release. The orde…
President Trump signed an executive order on Tuesday establishing a voluntary framework for government review of frontier AI models before public release, ending weeks of internal White House conflict over regulation. Th…
Polish Prime Minister Donald Tusk announced a "sovereignty test" for significant government technology purchases and annual IT independence reports, warning that Poland's dependency on foreign digital infrastructure dema…
Florida is suing OpenAI and CEO Sam Altman for allegedly marketing ChatGPT as safe while concealing known risks, including its role in a 2025 mass shooting at Florida State University and multiple suicides. Attorney Gene…
IBM and Red Hat launched Project Lightwell last week with a $5 billion commitment and 20,000 engineers to secure open-source software, while Chainguard CEO Dan Lorenc announced a $50 million, 100-engineer effort to build…
Agentic AI systems fail silently, wasting time and money without users noticing, according to a practitioner's analysis of production incidents from 2025. Data shows success rates drop from 54% for narrow single-task age…
A college professor discovered that 12 students from diverse backgrounds submitted nearly identical final essays written with AI, revealing a growing trend of young people outsourcing their internal experiences to artifi…
Datasette released version 0.1a0 of datasette-agent-micropython on June 2, 2026, enabling MicroPython execution within a WASM sandbox as a tool for Datasette Agent. The alpha release aims to allow the agent to generate a…
Xint, an AI-powered penetration testing tool, discovered an insecure direct object reference (IDOR) vulnerability in a healthcare client's system that exposed patients' protected health information. The vulnerability byp…
Microsoft released ASSERT, an open-source framework that uses AI to convert plain-language descriptions of intended behaviors into automated tests for evaluating application-specific AI systems. The tool generates test c…