Revolut launches a dedicated AI division and unveils PRAGMA, a proprietary foundation model built with NVIDIA, reporting a 65% fraud detection boost, 2.3x credit risk accuracy improvement, and 41% better personalized recommendations across 80 million customers in over 40 markets.
Vercel Connect is now generally available, replacing risky long-lived credential tokens with short-lived, auto-expiring tokens for AI apps, featuring 100+ connectors for services like Slack, GitHub, and Shopify, plus new RBAC controls, audit logs, and token observability across all plans.
Anthropic upgrades Claude Tag to read full Slack channels and proactively join conversations unprompted, marking a bold shift toward 'multiplayer AI' that acts as an organization-wide chief of staff, connecting enterprise data systems to pursue company goals — with security safeguards in place but pricing for the expanded feature still undecided.
Anthropic launches Enterprise Managed Auth, enabling Claude Team and Enterprise users to silently connect to MCP servers via their organization's existing SSO session, eliminating OAuth consent screens through a secure JWT bearer token exchange that automatically refreshes without user interaction.
Pi, an open-source AI agent toolkit, surges to 97K GitHub stars by offering unified multi-provider LLM support across OpenAI, Anthropic, and Google, complete with an agent runtime, interactive CLI, and robust supply-chain security through pinned dependencies and strict lifecycle script controls.
Google Cloud and Apple are joining forces to power Apple's Private Cloud Compute on Google Cloud infrastructure, using cutting-edge confidential computing technologies — including custom Titan chips, Intel TDX, and NVIDIA Blackwell GPUs — to deliver end-to-end AI security with full transparency through an open-source verification framework.
Nine people, including employees from Nvidia and Super Micro, are charged by Taiwanese prosecutors for smuggling banned AI servers containing high-end B300 GPUs to China, with 74 servers successfully reaching the country through Indonesia, Japan, and Hong Kong before authorities intercepted 56 more.
SpaceX and Nvidia are partnering to launch orbital AI data centers by Q4 2027, Thomson Reuters debuts a proprietary legal AI model outperforming top competitors, and Chinese state-linked hackers have more than doubled their cyberattacks after weaponizing DeepSeek's low-cost, low-guardrail capabilities.
Chinese state-affiliated hackers have more than doubled their cyberattack volume by integrating DeepSeek and other open-source AI models into their operations, using the technology to automate routine tasks and develop advanced malicious software, according to Taiwanese research firm TeamT5.
Anthropic launches a $35M Defender Advantage Fund and deploys Claude for enterprise code scanning, offering controlled outputs like patches and alerts rather than direct model access to prevent misuse, with timing aligned to incoming EU Cyber Resilience Act obligations.