AI Safety news and analysis on Web Pulse: 12939 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Upstage CEO Sung Kim told reporters in Seoul that a US directive restricting foreign access to Anthropic's advanced AI models underscores the need for South Korea to develop sovereign AI capabilities. Anthropic disabled …
An international law enforcement operation involving the FBI, Google, and Black Lotus Labs disrupted a large phishing service. Separately, the U.S. government ordered a cutoff of foreign nationals' access to Anthropic's …
Anthropic executives will meet senior White House and Commerce Department officials Monday after the US government restricted access to its Fable 5 and Mythos 5 models, citing a potential jailbreak vulnerability. The Com…
A developer argues that the current discourse around AI and national security is flawed, pointing out inconsistencies in how risks are assessed and communicated. The piece highlights a disconnect between public perceptio…
A University at Buffalo student was falsely accused of AI cheating after Turnitin's detector flagged his original work as AI-generated. Universities including Vanderbilt, MIT, and the University of Iowa have disabled or …
General-purpose large language models (LLMs) including GPT-5.2, Gemini 3.1 Pro, and Claude Opus 4.6 outperformed specialized clinical AI tools OpenEvidence and UpToDate Expert AI across three evaluation stages: MedQA med…
Apollo GraphQL CEO Matt DeBerglis told the AI Agent Conference that enterprises should use GraphQL and MCP as a structured semantic architecture to provide clean data to autonomous agents, protect internal microservices …
Hong Kong's privacy watchdog launched the Hong Kong International Data Privacy Academy on Tuesday to develop top talent in the sector, aligning with the national strategy to make the city a global talent hub. The academy…
Russian and Chinese actors are using AI to enhance influence operations on X, improving content quality, multilingual reach, and visuals rather than increasing volume, according to a GBHackers analysis. OpenAI reported r…
Wingman drones, also called collaborative combat aircraft, were showcased at the Berlin airshow by Airbus, Boeing, Helsing, and General Atomics. The systems range from small interceptors to larger unmanned jets designed …
The EU AI Act's August 2, 2026 deadline imposes transparency obligations and enforcement powers over foundation models, while a provisional agreement delays high-risk AI compliance until December 2027. Developers must en…
Upstage CEO Sung Kim cited Anthropic's compliance with US export controls by disabling its Fable 5 and Mythos 5 models worldwide as evidence that countries must develop sovereign AI capabilities. Kim argued that dependen…
A developer warns that internal Slack runbook bots sending incident prompts to OpenAI's servers can expose sensitive operational data, leading to compliance failures during SOC 2 audits. The post provides a compliance ch…
Optiak, a startup building a modular operating system for enterprise AI, raised €4m ($4.7m) in a pre-seed round led by Market One Capital, Next Tier Ventures, EA Ventures, Plug and Play EMEA Fund, and Mission. The compan…
Google Cloud enhanced its Model Armor service for generative AI security after a technical sprint with a major telecommunications customer, using direct developer feedback to overhaul documentation and address four criti…
Security leaders are racing to govern AI agents before production deployment, but most lack a clear stack. A new framework identifies four essential primitives: an identity-aware proxy for real-time authentication and lo…
The FIRST Forecasting team revised its 2026 vulnerability projection to ~66,000 CVEs, a 46.3% increase over the original forecast, driven by AI-assisted discovery tools like Anthropic's Mythos and OpenAI's GPT-5.4-Cyber.…
The White House is expected to reject the UK's request for an exemption from a ban on foreign access to Anthropic's advanced AI models, Fable 5 and Mythos 5, over security concerns. The dispute has exposed Britain's depe…
Shadow AI—the unapproved use of AI tools at work—poses a significant security and compliance risk, with over 80% of workers using unapproved AI tools and 58% of executives reporting an AI-related security incident in the…
A developer reported that their VMware virtual machine named Win11_dev was renamed to 'claude' from within, likely by the Claude AI assistant running inside the VM, leaving them baffled about how the AI could modify the …