RIS-Kernel Enables 64k Context LLMs to Run on CPU Using Sparse Attention

A new advancement in LLM technology, RIS-Kernel, demonstrates the capability to run large language models with a 64,000 token context on standard CPU architectures by utilizing sparse attention mechanisms. This breakthrough could significantly enhance the accessibility and efficiency of deploying powerful language models without the need for expensive GPU resources. The development represents a significant step towards optimizing computational resources for LLM applications.

Read Full Article

View All For This Day

AI AgentsZero-Trust+2

AI AgentsZero-TrustCybersecurityEnterprise Adoption

Anthropic Unveils Zero-Trust Architecture Framework for AI Agents

Anthropic has introduced a zero-trust architecture framework for AI agents, published on March 26, 2026, which addresses four critical threat vectors inadequately managed by traditional access controls. The framework comprises three implementation tiers: Foundation, Enterprise, and Advanced, each tailored to specific security needs, including isolation, audit trails, and real-time anomaly detection. Notably, it highlights that vulnerability-to-exploit timelines have drastically shortened, necessitating a reevaluation of existing security models. Anthropic asserts the importance of establishing robust security measures for AI agents before their widespread deployment, providing concrete architectural patterns rather than theoretical guidance. The release is expected to prompt enterprise security vendors to develop agent-specific zero-trust solutions and for Anthropic to incorporate these controls into its Claude API and MCP reference implementation.

Generative AICybersecurity+2

Generative AICybersecurityAI in MilitaryMalware

Iran's Military Leverages ChatGPT to Enhance Cyber Operations

Iran is utilizing Western AI models, including ChatGPT, to bolster its cyber operations by developing malware and conducting cyberattacks. This strategic use of generative AI technologies underscores the implications of advanced AI applications in military contexts. The integration of such tools is transforming Tehran's capabilities in the cyber warfare arena, raising concerns about the security and ethical ramifications of AI technology in military applications.

Financial TimesRead →

Generative AIStartup Funding+2

Generative AIStartup FundingProduct LaunchEnterprise Adoption

Investors Target Asian AI Companies Amid SpaceX and OpenAI Stock Offerings

Investors are actively seeking out companies poised to benefit from a surge in stock offerings in the United States, with a particular focus on the Asian supply chain. This trend is fueled by recent financial successes from prominent entities like SpaceX and OpenAI, which have sparked increased interest in the next wave of AI winners in Asia. The strategic shift aims to capitalize on emerging opportunities within the region's technology landscape.

Bloomberg TechnologyRead →

AICloud Computing+2

AICloud ComputingRegulationChina

China Urges Tech Giants to Shift Focus from Price Wars to AI Investment

A recent commentary from a top Communist Party journal has directed China's major tech platforms to cease price wars and instead invest in artificial intelligence (AI) and cloud computing. This shift indicates a move towards regulatory stabilization after years of scrutiny, encouraging companies such as Alibaba, Meituan, and PDD Holdings to prioritize value over aggressive subsidy competition. The commentary emphasizes the need for stronger oversight of algorithms, data usage, and consumer protection while signaling a transition from strict regulatory crackdowns to a more calibrated approach to governance. The call for increased investment in strategic technologies aligns with China's broader national strategy to lead in the AI sector, amid ongoing pressure on companies to adapt to new compliance and operational constraints.

The Next Web AIRead →

Generative AIHuman Rights+2

Generative AIHuman RightsPrivacyEthics

Amnesty International Calls for Ban on Generative AI Due to Human Rights Violations

A recent briefing by Amnesty International highlights the human rights costs associated with standalone generative AI systems, which rely on unlawful web scraping. The organization argues that these technologies violate international human rights law (IHRL) through their design, development, and deployment practices. The report indicates that such systems compromise privacy rights, promote discrimination, and threaten freedom of expression. Amnesty International asserts that the fundamental incompatibility of these AI systems with IHRL necessitates their prohibition.

Hacker NewsRead →

Data CentersAI Infrastructure+2

Data CentersAI InfrastructureCommunity ProtestsEnvironmental Impact

Utah Governor Enacts New Standards for Data Centers Amid Community Backlash Over 9 GW Project

Utah Governor Spencer Cox has signed an executive order establishing new standards for data center development following community protests against the 9 gigawatt Stratos Project, backed by investor Kevin O’Leary. The order, effective immediately, introduces eight principles focused on water resources, air quality, and wildlife protection, mandating that Stratos developers apply for new permits for each planned expansion rather than receiving blanket approval. Local opposition has been significant, with residents expressing concerns over environmental impacts, while supporters advocate for job creation and economic growth. The controversy highlights the growing tension between AI infrastructure demands and local community interests, as similar issues arise nationwide.