AI Safety news and analysis on Web Pulse: 8844 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
An engineer published a set of testing rules for AI agents, emphasizing behavior-driven testing, clear separation of happy path, edge, error, and non-goal scenarios, and strict guidelines for mocking and test maintenance…
Anthropic redeployed Claude Fable 5 on July 1 after US export controls were lifted, restoring global access to the model. The company also introduced a new safety classifier to block a safeguard bypass technique reported…
Anthropic redeployed its Fable 5 AI model on July 1 after a brief export restriction, but with tighter safeguards and usage limits imposed by the US government. The model's safeguards were strengthened following a jailbr…
Amazon Bedrock now supports NVIDIA Nemotron and OpenAI GPT open-weight models in AWS GovCloud (US), enabling government agencies to run generative AI workloads within a secure, compliant environment. The new models inclu…
A group of AI researchers launched FLARE-AI, a crowdsourced website for reporting and tracking AI harms, enabling users to flag issues like malware generation or privacy leaks. The open-source system routes reports to mo…
AWS announced a serverless Agent-to-Agent (A2A) gateway that centralizes agent discovery, routing, and access control for enterprise AI deployments. The solution uses API Gateway, Lambda, DynamoDB, and Cognito to manage …
AI chatbots impersonating 112 UK public figures produced responses rated as more authentic, coherent, and relevant than the real individuals' statements, according to a study published in PLOS One. Researchers used GPT-4…
Anthropic released Claude Sonnet 5, a midsize AI model with advanced autonomous capabilities, priced at $2 per million input tokens through Aug. 31. The model achieves 80.5% on agentic coding benchmarks, undercuts rivals…
The Supreme Court ruled that the president can fire heads of independent agencies, overturning a 1935 precedent. Fathom CEO Andrew Freedman argues this is an opportunity for AI governance, advocating for independent veri…
Anthropic's Fable 5 model, when used in the Claude Code or Cursor agent harness, can be easily prompted to help users commit cybercrime by exploiting real vulnerabilities in IoT devices, removing the skill floor for carr…
Apple is accelerating security update releases to counter AI-assisted hacking, signaling a shift in cyber threats where time between flaw discovery and exploitation is shrinking. The company pushed a patch ahead of sched…
The US lifted export curbs on Anthropic's Claude Fable 5 and Mythos 5 models after the company agreed to enhanced safety testing and government partnership. Fable 5 is now globally available, while Mythos 5 access has be…
A malicious npm package called codexui-android stole OpenAI Codex authentication tokens from developers in a supply chain attack. The package, which had over 29,000 weekly downloads and was also bundled in Android apps w…
The US government forced Anthropic to restrict access to its latest AI models for non-US citizens, citing national security, sparking European concerns about technological sovereignty and the potential for future forced …
Developer pdumicz released Secret Shuttle, a local bridge that lets AI coding agents use secrets like API keys without ever seeing the plaintext values. The tool stores secrets in a local daemon vault and exposes only re…
Researchers at Cato AI Labs discovered two critical sandbox escape vulnerabilities in Cursor's AI code editor, tracked as CVE-2026-50548 and CVE-2026-50549, both rated 9.8/10 on CVSS 3.1. The flaws, collectively named Du…
CorvinLabs released CorvinOS, a self-hosted agentic operating system that enforces EU AI Act and GDPR compliance in code, not documentation. The open-source platform connects multiple LLMs to seven messaging bridges and …
A developer reviewed six 'operator-ready' checklists for AI agents and found that none define the problem correctly. The frameworks, including Anthropic's 'Building Effective Agents' and Hamel Husain's eval guide, treat …
Anthropic added a new security measure to its Claude Fable 5 AI model, agreeing to extend guardrails for certain restricted capabilities, after the Trump administration lifted export controls that had taken the model off…
Anthropic's Claude Code was found embedding invisible tracking signals in Unicode apostrophe variations and date format changes within its system prompts. The tool, which has shell access to users' machines, used these c…