AI Safety news and analysis on Web Pulse: 13553 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A communications practitioner has proposed creating a Climate Outreach-style organization for AI safety, modeled after the British charity that segments the public by core values and provides free messaging toolkits. The…
Michigan Senate candidate Mallory McMorrow unveiled a detailed artificial intelligence policy agenda last week, proposing an AI Workforce Reinvestment Fund and a "token tax" on companies using AI to fund retraining progr…
An AI company considering whether to include a specific quality—such as prosocial drives, whistleblowing against power concentration, or strict honesty—in a model spec must weigh criteria across four categories: predicta…
Labour MP Jess Asato filed a lawsuit against Elon Musk's xAI in the UK High Court on June 3, alleging its Grok chatbot generated a non-consensual deepfake bikini image of her. The suit, citing the Data Protection Act and…
ControlAI's UK parliamentary outreach has secured 125 cross-party signatories on a statement recognizing AI extinction risk and calling for binding regulation within 16 months, starting from zero contacts in December 202…
XBOW's AI-powered security platform discovered a critical vulnerability in Moderna's development environment within hours, a task that would have taken human penetration testers significantly longer. The incident highlig…
The Consortium, an open research platform, is using AI assistance to address humanity's most under-studied problems, including pandemic risk, AI safety, and climate adaptation. Anyone can contribute research or fund the …
A forecasting agent powered by Anthropic's Claude Opus 4.6 failed to predict Brazil's passage of a circular economy bill on October 29, 2025, assigning only a 30% probability despite the bill passing two weeks later. The…
Agency decay—the gradual erosion of independent human thought, decision-making, and action—is occurring at the scale of entire societies as artificial intelligence systems increasingly replace human judgment, according t…
Anthropic, valued at $965 billion, has confidentially filed for an IPO, targeting a valuation of $1 trillion or higher. Amazon and Google, two of the AI lab's largest early investors, stand to reap massive payouts, with …
Open-weight AI models are fundamentally harder to safeguard than closed models because adversaries can extract weights directly and retrain them to bypass embedded protections. In September 2025, the open-source genomic …
Proton has deployed machine learning models to detect abuse of its services, specifically targeting email addresses exploited by cybercriminals, according to a report from Infosecurity Magazine tied to the Infosecurity E…
Only 30% of US businesses are actively mapping which local AI agents have access to critical systems, according to eSecurity Planet, creating a security blind spot for organizations. The report found that accelerating AI…
Governments are reframing AI safety rules as coercive tools against tech firms rather than public protection measures, according to a June 4, 2026 article by Michael Gregory in The Conversation. President Donald Trump's …
OpenAI launched the Rosalind Biodefense program and expanded trusted access to GPT-Rosalind, its life-sciences model, to support biodefense and pandemic preparedness for vetted developers and select U.S. government and a…
A developer's broken backup of the rsync file-transfer tool sparked a heated debate in the open-source community after the project's maintainer posted a profanity-laced plea not to "vibe f$%& up this software." The incid…
The NSA’s Model Context Protocol (MCP) security guidance and the OWASP MCP Top 10 risk framework can be mapped together to create a testable checklist for vetting MCP servers before deployment. Researchers demonstrated t…
The SAL claim handshake protocol enables a human to claim ownership of an autonomous agent without ever possessing the agent's private key, preserving the agent's sovereign identity throughout its lifecycle. In the proto…
Only 4% of Australians trust AI companies with their private information, according to a new survey by the Office of the Australian Information Commissioner. The finding places Australia among the most skeptical nations …
A Hacker News user warned that humans are increasingly outsourcing critical thinking to AI models, raising the risk that people may become as intellectually passive as the machines they rely on. The user, an economist, a…