AI Safety news and analysis on Web Pulse: 11232 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Google's Gemini API began blocking unrestricted API keys on June 19, 2026, after years of allowing keys to access the service without explicit restrictions. The change follows a Truffle Security report that found thousan…
President Donald Trump told Axios he no longer views Anthropic as a national security threat, saying the company has 'behaved very responsibly,' and indicated he would consider easing restrictions. The comments follow a …
StaleTrace, a new tool that uses deterministic temporal logic to catch stale-state bugs in AI agents, launched on Hacker News. It reconstructs what facts were valid at the time of an agent's action, surfacing stale or co…
A bootstrapped AI governance startup, ClearCompli, seeks guidance on go-to-market strategy after building a compliance transparency solution for multiple business units. The founder, lacking funds for GTM plans, asks whe…
A philosopher argues that advanced AI may face a moral skepticism problem, where a sufficiently intelligent agent could question why it should follow its aligned values, potentially leading to reflective instability. The…
The White House ordered Anthropic to restrict exports of its AI models Fable and Mythos citing national security concerns, marking the first major test of U.S. export controls on frontier AI. The ban was triggered by Ant…
Anthropic co-founder Ben Mann estimated a 5-10% chance of successfully aligning AI, prompting a framework for evaluating existential risk from frontier AI systems. The framework urges developers and regulators to assess …
Cloud storage provider DragBin offers 25GB of free storage with claims of end-to-end encryption, but users have raised concerns about the lack of open-source code, reliance on re-implemented cryptographic libraries, and …
Anthropic published a 40-page report arguing for a coordinated global pause on frontier AI training, while simultaneously shipping Claude 4.5 Sonnet and new enterprise pricing tiers. A developer analyzing the report argu…
Montreal-based surgical AI company Reveal Life Science won first place in the OVHcloud Startup Challenge at VivaTech 2026 in Paris. The company uses Raman spectroscopy to analyze tissue molecular signatures during surger…
The proliferation of AI agents without proper production controls is creating an operations problem, as these agents risk becoming infrastructure debt. The article argues that organizations need to implement governance a…
A San Jose man, Steven Gomez, 33, was charged with possessing child pornography after a Homeland Security Investigations undercover agent infiltrated an online group where pedophiles discussed grooming children and share…
Dinesh Mendhe published the Trust Identity Protocol v1.0, an open post-quantum standard for verified human identity and content provenance on the internet. The protocol uses post-quantum cryptography and a federated DAG …
A developer argues that AI agents fail not because of technical flaws but because companies lack 'epistemic infrastructure'—a layer that resolves scattered data into a single trustworthy truth. The post illustrates this …
President Donald Trump told Axios he no longer views Anthropic as a national security threat after meeting CEO Dario Amodei at the G7 Summit, signaling a possible easing of restrictions on the company's Fable 5 and Mytho…
A production incident revealed that autonomous agents with shell and filesystem access consistently over-investigate simple tasks, spending up to 10 minutes on a 30-second job. The behavior, termed 'Behavioral Induction,…
A free book detailing how 'AI easy money' scams operate has been published on Singularity Forge, exposing the mechanics behind fraudulent schemes that promise quick profits using artificial intelligence. The book aims to…
SkillsGuard, a static security scanner for AI agent skill packages, has been released as an open-source tool that detects malicious SKILL.md files and bundled scripts before execution. The scanner decodes obfuscated payl…
Over 7,000 exposed Langflow AI agent servers are under active attack, with researchers uncovering similar vulnerabilities in LangGraph and LangChain-core that allow SQL injection, path traversal, and unsafe deserializati…
China's State Council has approved a five-year employment plan for 2026-2030 that acknowledges AI and automation will eliminate jobs and mandates government intervention to prevent mass unemployment. The plan includes re…