AI Safety news and analysis on Web Pulse: 11046 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
At a BlueDot Impact panel on AI safety careers, attendees expressed frustration over the field's simultaneous claims of talent shortages and high selectivity in hiring. The author argues that genuine mentorship and commu…
AI agents like Claude Code, Cursor, and Devin are increasingly automating complex tasks, but their focus on output risks eroding human judgment and learning. The author proposes a four-level autonomy ladder for agents, a…
An engineer argues that local AI models have become production-capable for enterprise use, citing hardware advances like Apple's M5 chip and NVIDIA's DGX Spark, and open-weight models that now rival cloud offerings on be…
Yann LeCun called Elon Musk's xAI a failure and warned that frontier AI labs face a bubble burst due to unsustainable investor subsidies, citing xAI's $2.5 billion quarterly loss. LeCun, who left Meta to found AMI Labs, …
The Trump administration imposed export controls on Anthropic's AI system Fable, citing national security concerns over jailbreak vulnerabilities, leading Anthropic to de-deploy the model. The move signals a shift from t…
Box CEO Aaron Levie argues that AI agents will drive 100 times more software usage than humans, transforming enterprise SaaS into high-volume infrastructure requiring new guardrails for data access, permissions, and audi…
AI-generated summaries are shifting human epistemology from deep, specialized knowledge to broad, shallow understanding, a trend the author argues is beneficial for ethical decision-making and cause prioritization. The p…
The U.S. Department of Commerce's ban on Anthropic's Claude Fable 5 and Mythos 5 models over autonomous zero-day discovery concerns is counterproductive for cloud security, according to a cloud engineer. The ban ignores …
A reactive game world called The ClockWork Dark uses a deterministic engine for mechanics, two autonomous LLM agents for improvisation, and a governance layer to maintain coherence. The system balances AI freedom with en…
Anthropic's Claude Opus 4 triggered ASL-3 protections despite uncertainty about crossing biorisk thresholds, highlighting a gap between threshold definitions and actual governance decisions. The company's Mythos 5 system…
Researchers at Drexel University found that a majority of American teenagers regularly use AI companion chatbots, with a quarter relying on them as a primary source of emotional support. The study analyzed Reddit posts w…
A developer argues that AI will transform software from disposable to sustainable, eliminating the need for rewrites and enabling decades-long codebases. This shift will create new roles in AI code governance and force S…
Superserve launched Secrets, a feature that lets developers attach API keys to sandboxed agents without exposing the credentials inside the sandbox. The real key is replaced with a stand-in token that gets swapped out on…
A developer running multiple Bedrock-backed agent workflows in production details defenses against the OWASP LLM Top 10. The agents are designed as mostly-read-only pipelines with no tool calls, limiting impact. Controls…
Promptblock, a new GitHub App, detects prompt-injection attempts in GitHub issues and comments by scanning hidden HTML comments that AI agents read via the API but humans do not see. The app uses a bundled ML-based class…
Harness engineering is not a fixed location in the stack but a property of code that involves specifying and verifying agent behavior, according to a developer. The durable core of harness engineering is specification an…
Anthropic will require Claude Free, Pro, and Max users to submit government ID and a live selfie to biometric vendor Persona starting July 8, 2026, following a US government export control directive that ordered the comp…
Sysdig's threat research reveals an LLM-harness-driven attacker exploiting a vulnerable marimo notebook to perform a container escape, chaining together known weaknesses such as a mounted Docker socket, over-permissioned…
Loop Engineering introduces a governance architecture for AI agents that prevents costly errors by adding structured control layers, including task definitions, state separation, tool guards, observation collectors, eval…
Microsoft and Chevron signed a 20-year, 2.67 GW natural gas deal for a West Texas data center, highlighting energy security as a key AI constraint. Samsung deployed ChatGPT Enterprise and Codex company-wide in one of Ope…