Latest AI (Artificial Intelligence) News
Anthropic Restricts Third-Party Claude Usage to Prevent Cyberattacks
Anthropic is limiting access to its new AI model amid concerns over cyberattacks, as enterprises seek to manage AI chaos. The company is prioritizing safety by restricting third-party usage of Claude. This follows reports of AI models being used in cyber espionage.
Google Releases Gemma 4 Family of Open-Source Models
Google has launched the Gemma 4 family, a new set of open-source AI models highlighted in recent updates. These models aim to advance accessible AI development. They compete in the growing landscape of open-source AI tools.
Meta Unveils Muse Spark Reasoning Model
Meta introduced Muse Spark, a new reasoning AI model, re-entering the competitive AI race. The model supports the shift toward agentic AI alongside leaders like OpenAI and Anthropic. Meta plans significant investments in AI infrastructure.
DIA Launches Digital Modernization Accelerator for Centralized AI
The US Defense Intelligence Agency has centralized its AI efforts with the new Digital Modernization Accelerator, evolving from Task Force Sabre. It supports Combatant Commands and deploys ChatDIA on classified networks. Plans include agentic AI for semi-autonomous assistance.
Cursor 3.0 Major Update Boosts Agentic AI Capabilities
Cursor releases version 3.0, a significant update enhancing AI coding and agentic workflows. It features alongside tools like OpenClaw in the agentic AI surge. This positions it as a key player in developer AI tools.
AI Browser Extensions Pose High Enterprise Security Risks
AI browser extensions are a new, unguarded AI consumption channel with 60% more vulnerabilities than average. They bypass DLP controls, access sensitive data, and frequently change permissions. Enterprises must monitor this growing threat surface.
Anthropic Acquires Coefficient Bio Amid AI Expansion
Anthropic has acquired Coefficient Bio, expanding into biotech applications of AI. This move aligns with broader AI advancements in specialized domains. It reflects ongoing consolidation in the AI sector.
AI-Powered Cyber Threats Evolve with Model Hijacking
Adversaries use LLMs for 80-90% of cyber operations, including prompt injection for model hijacking. Defenders counter with human-guided AI agents for faster threat detection. Core security principles remain essential against AI threats.
Bank CEOs Discuss Anthropic AI Risks with Regulators
Federal Reserve's Powell and others summon bank CEOs to address Anthropic's AI developments amid Wall Street concerns. This highlights growing financial sector scrutiny of AI disruptions. Software shares fall on AI-related risks.
Alibaba Releases Open-Source Wan 2.1 AI Video Model
Alibaba launched Wan 2.1, an open-source AI video generation model, advancing multimedia AI. It joins global efforts in accessible generative tools. This boosts competition in AI video tech.
Meta Commits $21B More to CoreWeave AI Infrastructure
Meta announced an additional $21 billion investment in CoreWeave’s AI infrastructure to fuel its AI ambitions. This underscores massive capital flowing into AI compute. It supports models like Muse Spark.
Anthropic Previews Mythos Model, Held Back for Safety
Anthropic previews its unreleased Mythos model but delays full release due to safety concerns. This cautious approach addresses risks in advanced AI deployment. It follows disputes like Anthropic vs. Pentagon.