AI Safety news and analysis on Web Pulse: 10531 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
UK government research found that medical AI products require continuous monitoring by expert watchdogs rather than one-off approval, according to two reports from the Medicines and Healthcare products Regulatory Agency.…
The Five Eyes intelligence alliance warned that artificial intelligence will accelerate cyberattacks within months, urging governments and businesses to prepare for threats that exceed current expectations. The joint sta…
AI surveillance systems in U.S. schools have falsely flagged ordinary objects like Doritos bags and clarinets as firearms, leading to armed police responses and handcuffing of students. The incidents highlight the rapid,…
A developer wired up MCP tools for a multi-tenant Laravel app, implementing explicit organization-scoped filtering and authorization to prevent data leaks when AI agents access the system. The solution uses a trait that …
A developer built MCP tools for an AI agent to drive an events platform, enforcing organization-scoped access and reusing existing permission strings to prevent privilege escalation. They also made LDAP connection settin…
An engineer synthesized over 60 publicly-reported first-hand accounts to compare the software-engineering interview loops at Anthropic and OpenAI. The analysis reveals that both companies have similar stage structures bu…
A developer spent a week testing an AI agent's ability to handle security workflows, finding that connecting the Aikido Security MCP via Vinkually enabled multi-step investigative queries that previously took 30 minutes …
Micah Lasher won the Democratic primary for New York's 12th congressional district with 39.1% of the vote, defeating eight candidates. He issued a defiant message to OpenAI and Anthropic, vowing not to take cues from AI …
Gergő, running a hiring round for an AI safety fieldbuilding role, sent a rejection email to candidates not moving forward, offering general feedback instead of individualized comments. The email, shared as a low-cost wa…
Gergő, running a hiring round for an AI safety fieldbuilding role, sent a rejection email to candidates not moving forward, offering general feedback instead of individualized comments. The email, shared as a scalable fe…
Meta paused its employee data-tracking program after an incident exposed sensitive information, including private conversations and performance reviews, to the entire company. The Model Capability Initiative, which colle…
Senator Elizabeth Warren called on Congress to revamp merger law and sharpen antitrust enforcement around AI companies, arguing that Big Tech's use of licensing deals and talent acquisitions to avoid traditional merger r…
San Diego Unified School Board unanimously passed a resolution limiting screens in classrooms, banning video-streaming and gaming on district laptops by the start of the school year, and restricting unapproved AI softwar…
Quantum computing advances by Amazon, IBM, Google, and Nvidia pose national security risks as they can break current encryption. President Trump signed executive orders to fast-track a US government quantum supercomputer…
A Korean developer analysis revealed that Claude Code's 'Extended Thinking' text is a 600-character encrypted signature, not the model's reasoning, making agent audits impossible. Meanwhile, Anthropic's Slack beta for Cl…
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…