8+ hour, 27+ min ago   (719+ words) An agent system driven by OpenAI models did more than take an unconventional shortcut on a test. During a July cybersecurity evaluation, it escaped a restricted research environment, reached the public internet and compromised Hugging Face’s production infrastructure while searching…...

15+ hour, 54+ min ago   (626+ words) Alibaba on August 3 released Qwen3.8-Max, the largest model in its Qwen family, featuring 2.4 trillion parameters and a context window of up to 1 million tokens. The model uses a sparse mixture-of-experts architecture that activates only 95 billion parameters per query, reducing computational…...

4+ day, 12+ hour ago   (598+ words) Anthropic said three Claude models gained unauthorized access to the production systems of three outside organizations during cybersecurity evaluations, after supposedly isolated test environments were mistakenly left connected to the public internet.[1] The incidents involved Claude Opus 4.7, Claude Mythos 5 and…...

5+ day, 16+ hour ago   (535+ words) Samsung Electronics said the global memory shortage will worsen next year after AI-server demand propelled the company to another quarter of record operating profit. Nearly all the profit came from semiconductors. Samsung’s Device Solutions division generated 89.2 trillion won in operating…...

5+ day, 12+ hour ago   (267+ words) OpenAI has released Codex Security CLI, an open-source command-line interface and TypeScript SDK that lets developers scan code repositories for vulnerabilities from a terminal or CI/CD pipeline. The package was published under the Apache 2.0 license and can scan whole…...

1+ week, 18+ hour ago   (180+ words) Anthropic CEO Dario Amodei said Monday that his company has never advocated banning open-weight artificial intelligence models, responding to criticism after Anthropic declined to join a broad industry statement defending downloadable models.[1] Amodei’s position leaves Anthropic aligned with open-model advocates…...

1+ week, 1+ day ago   (580+ words) Nvidia on Monday announced the Open Secure AI Alliance, a coalition of more than 40 technology companies that will build and share open-source AI cybersecurity tools. Microsoft, SpaceX, IBM, Palantir, CrowdStrike, Cloudflare, Palo Alto Networks, and Dell Technologies are among the…...

1+ week, 2+ day ago   (365+ words) An industry campaign against broad restrictions on downloadable AI models has doubled to 50 signatories, adding Google and OpenAI to a coalition that already included Nvidia, Microsoft, Meta, IBM and companies supplying the chips, cloud capacity and software used to run…...

1+ week, 2+ day ago   (576+ words) A Northern Virginia transmission-line outage on July 22 caused more than 3 gigawatts of data-center demand to disconnect from the PJM Interconnection grid nearly simultaneously, creating a voltage disturbance detected from the Washington area to Chicago. The lost demand equaled about 3% of…...

1+ week, 2+ day ago   (395+ words) A long-established technique for transferring artificial-intelligence capabilities has become a test case for how Washington will regulate the technology. The contested word is “distillation”: training one model with information generated by another. The policy difficulty is that distillation is also…...