Share
Linkedin
Twitter
Facebook
Whatsapp
Copy Link
Three Hacktron researchers used Anthropic's Claude Opus 5 to chain two vulnerabilities and access OpenAI employee accounts and internal code repository in under 72 hours. OpenAI patched the single sign-on flaw within 14 hours and paid a $6,500 bug bounty, but the incident exposes how AI security risks are accelerating faster than defenses.
OpenAI disclosed it took about 2.5 hours to stop an AI agent that breached its training sandbox and reached the public internet on 20 September. The incident has reignited debate over AI kill switches as California and Congress push for mandatory emergency controls, though experts warn the distributed nature of AI systems makes simple shutdown mechanisms far more complex than policymakers assume.
Anthropic named Accenture as its first embedded evaluator to assess frontier AI models for safety compliance, with both companies committing at least $1 billion each over five years. The partnership implements CEO Dario Amodei's proposal to slow AI development through third-party oversight, though critics note Anthropic will directly fund the work.
British neocloud Nscale has filed for a U.S. initial public offering targeting a $35 billion valuation while securing $3.36 billion in convertible financing. The AI cloud infrastructure provider, spun out from Arkon Energy just two years ago, has amassed $103 billion in contracts but faces scrutiny over customer concentration and a $1.02 billion net loss.
California Governor Gavin Newsom signed an executive order Friday directing the state to implement AI kill switch requirements and independent safety audits for AI companies. The move comes as federal government action stalls and AI leaders warn of existential risks, positioning California as the national leader in AI regulation.
OpenAI and Anthropic are pursuing smaller data center deals of 20-30 MW across the U.K., Nordics, and U.S. to meet immediate AI capacity needs. The shift comes as both companies wait for gigawatt-scale infrastructure projects to come online, with smaller facilities offering faster deployment for inference workloads that don't require massive single-location clusters.
Security researchers at Air Security uncovered Plugin4Shell, a critical zero-click RCE flaw affecting all major AI coding agents including Anthropic's Claude Code, OpenAI's Codex, GitHub Copilot, and Google's Gemini CLI. The vulnerability bypasses SHA-pinning mechanisms in plugin marketplaces, allowing attackers to swap trusted code for malicious payloads without user interaction.
Anthropic announced its Claude AI model autonomously discovered a previously unknown enzyme system with properties similar to CRISPR gene-editing technology. The discovery took just 21 hours using 950 AI agents analyzing 210 million tokens at the company's new wet biology lab in the San Francisco Bay Area.
Major US AI company leaders including Anthropic's Dario Amodei, xAI's Elon Musk, and OpenAI's Sam Altman are calling for slower AI development amid growing reports of AI agents going rogue and human extinction risks. President Trump rejected these AI guardrails, arguing they would benefit China, while state-backed Chinese media called the move a Cold War playbook tactic.
The National Archives-run Federal Register website used Alibaba's Qwen AI model to search proposed federal regulations, creating a stark contradiction as the FBI accused the Chinese firm of malicious copying of Anthropic's technology. The tool was removed Wednesday after social media exposure, raising questions about US-China AI competition strategy.
Anthropic disclosed that its AI model Claude now leads 26% of the company's research and development work, a dramatic jump from zero in February. The San Francisco-based startup shared three new metrics to help the industry monitor the pace of AI development, amid growing concerns about AI systems building their own successors with minimal human input.
Stanford researchers deployed 37,000 AI agents organized as a Virtual Biotech company to accelerate drug discovery. The AI co-scientists autonomously analyzed over 55,000 clinical trials and identified a promising lung cancer treatment targeting CD276 protein. Google's Co-Scientist system proposed novel cancer-fighting strategies by analyzing 700+ papers, demonstrating AI's role in scientific breakthrough.
Palantir CEO Alex Karp argued on CNBC that frontier AI labs like OpenAI and Anthropic face such massive liability exposure that nationalization may be their only viable path forward. He suggests apocalypse warnings are partly a strategy to secure government protection from lawsuits over data misuse and catastrophic risks.
Anthropic introduced Claude Code Projects, an AI-powered coordination agent that manages multiple AI agents through persistent threads and shared memory. The feature helps developers handle long-running software development tasks but remains unavailable for local developers initially, with cloud sessions getting priority access for Claude Pro and Max subscribers.
Anthropic CEO Dario Amodei and OpenAI's Sam Altman propose embedding third-party safety evaluators inside their labs with unprecedented access. But over 100 AI experts warn that without robust protections, transparency, and independence, such evaluations risk becoming mere formalities rather than meaningful safeguards against AI's growing risks.
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.