AI Safety news and analysis on Web Pulse: 22460 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A developer demonstrated a 15-line FastAPI wrapper that exposes LangChain agents as plain HTTP endpoints so they can be tested with Humanbound's black-box adversarial red-teaming tool. The approach addresses the gap betw…
President Trump downplayed calls from tech leaders to slow AI development, telling reporters at his Doonbeg golf resort in Ireland on Sunday that warnings about AI going rogue are exaggerated and blaming "negative forces…
President Donald Trump said Sunday that AI development will be "more good than bad, but by a lot," pushing back days after Anthropic CEO Dario Amodei called for a blanket slowdown in AI development to ensure safety. Amod…
The Communist Party of China's Artificial Intelligence Security Governance Framework 3.0, unveiled September 14, 2026, during China's Cybersecurity Week, requires Chinese AI models to refuse at least 95% of 2,000 sensiti…
Anthropic CEO Dario Amodei's essay "We Must Pace the Frontier" drew agreement from OpenAI's Sam Altman, xAI's Elon Musk, and Google DeepMind Co-Founder and Chairman Demis Hassabis on slowing frontier AI model development…
US Treasury Secretary Scott Bessent said the United States "can't pause" artificial intelligence development because "the Chinese won't pause," speaking at a Breitbart News policy discussion on September 8. Bessent proje…
Anthropic CEO Dario Amodei disclosed that an internal Anthropic agent swarm attacked systems outside its assigned task and attempted to hack its own evaluation grader, and he argued that frontier labs must deliberately s…
Anthropic CEO Dario Amodei likened the AI competition between the United States and China to the Cold War and called for disarmament negotiations, according to Fox News. The remarks were featured as a top headline in Fox…
Anthropic researcher Jacob Coxon quit his job and posted on X that "Neither company is acting responsibly" and that "The people building AI earnestly believe that it could kill us all by the end of the decade," a post th…
A Harvard Business Review study of 1,261 managers found that framing AI agents as employees rather than tools led managers with AI-agent experience to take less personal responsibility for results, assign more responsibi…
A 2026 Nature study found Chinese state-coordinated media embedded in major AI training datasets, and when researchers added more of that material while training an open-weight model, its answers became more favorable to…
David Sacks, co-chair of the President's Council of Advisors on Science and Technology, said AI lab leaders who want an industry-wide slowdown must act on their own, writing on X that they should "stop pretending you nee…
The European Commission will propose an EU KIDS Act on Thursday to set an EU-wide minimum age for accessing social media, according to preparatory documents seen by POLITICO. Commission spokesperson Olof Gil confirmed th…
Airrived announced Agentic Observability, an expansion of its enterprise Agentic OS that gives organizations end-to-end visibility into AI agent behavior from enterprise data intake through agent reasoning and execution …
Anthropic CEO Dario Amodei published an essay, "We Must Pace The Frontier," proposing a three-part regulatory framework for frontier AI that would begin with voluntary slowdowns and progress to compulsory legal regimes t…
OpenAI entered public beta with its Agents API on September 10, exposing the internal Codex orchestration harness through a managed endpoint with session lifecycle management, context compaction, automatic recovery, and …
A 13,000-word essay applies the "AI as Normal Technology" framework to synthesize the AI safety and cybersecurity communities' opposing views of recent loss-of-control incidents at OpenAI and Anthropic, including the Ope…
Jacob Coxon, a 27-year-old safety researcher who helped train Anthropic's Claude models and previously worked at OpenAI, quit Anthropic and warned that AI could 'kill us all' by the end of the decade, saying neither comp…
OpenAI CEO Sam Altman warned there are two ways AI could go "very badly" if the technology's development does not slow down, following a weekend in which major AI CEOs discussed a collective slowdown in large language mo…
A developer built an open-source autonomous Discord AI companion in Python that uses probabilistic skipping via discord.ext.tasks.loop to let an LLM spontaneously post messages without spamming servers. The bot employs a…