Anthropic and the Bill & Melinda Gates Foundation have announced a four-year, $200 million partnership, with Anthropic contributing $100 million in technical support and Claude usage credits, and the Gates Foundation investing $100 million in funding and expert guidance. The collaboration focuses on three key areas: global health, education, and economic mobility. In health, it aims to improve healthcare services in low- and middle-income countries and accelerate vaccine development and modeling for diseases such as malaria and tuberculosis. In education, it will develop K-12 math tutoring and career guidance tools targeting the U.S., sub-Saharan Africa, and India. For economic mobility, it will enhance agricultural productivity for smallholder farmers and support skills certification systems.

OpenAI confirmed that its internal development environment was compromised in a supply chain attack due to a tampered open-source npm package, TanStack, by the TeamPCP ransomware group. The breach infected two employee devices and allowed attackers brief access to certain internal code repositories, including code signing certificates for macOS, Windows, and iOS products. OpenAI has proactively rotated the certificates and re-signed affected applications. macOS users must update their apps by June 12, or older versions will cease to function. The company emphasized that customer data, production systems, and core intellectual property were unaffected. This incident is part of a broader campaign, with Mistral AI and Microsoft also reporting similar attacks.

OpenAI has released an update to the ChatGPT mobile app, bringing Codex functionality to iPhone, iPad, and Android devices. Users can now remotely access Codex running on a Mac, enabling cross-device collaboration including approving tasks, initiating new commands, viewing terminal output and code diffs, and switching models—all synchronized in real time. Setup requires scanning a QR code generated on the Mac using the mobile device. The feature is currently available as a preview in all supported regions, with remote control for Windows coming soon. On the same day, OpenAI also launched new features for Codex, including Hooks and programmatic access tokens.

Figure AI completed an 8-hour fully autonomous factory shift demonstration on May 13, during which the Helix-02 robot sorted barcode-labeled packages at near-human speed without any human intervention. The system runs on a proprietary two-layer AI architecture: the S2 vision-language model handles task planning, while the S1 visual-motor policy controls real-time movements across 35 joints—all computed locally on the robot. The Figure 03 robot, equipped with high-sensitivity fingertip sensors, has already achieved coordinated complex tasks such as bedroom tidying with two robots working together. The company has delivered over 350 third-generation robots, and this eight-hour continuous operation marks a critical step toward practical industrial deployment of humanoid robots.

xAI has released an early test version of Grok Build, an agent-based command-line tool for coding, application development, and workflow automation, available to SuperGrok Heavy subscribers. Grok Build supports a fully interactive CLI interface, accepts mouse input, and manages multiple sub-agents, with agents capable of running sub-agents to verify their own work quality. Elon Musk personally promoted the tool on social media, encouraging user trials. This marks xAI’s first major move into the AI coding tools space, positioning it in direct competition with OpenAI Codex and Anthropic Claude Code.

PwC and Anthropic have announced an expanded strategic alliance to advance agent-based AI enterprise applications using Claude. PwC will deploy Claude Code and Claude Cowork, initially rolling out in the U.S. and expanding globally to hundreds of thousands of professionals, with plans to train and certify 30,000 U.S. employees. Production-level deployments are already underway in insurance underwriting, mainframe modernization, HR transformation, and cybersecurity, achieving up to 70% improvement in client delivery efficiency. Insurance underwriting cycles have been reduced from 10 weeks to 10 days, and security response times have dropped from hours to minutes. Healthcare leader Advocate Health is advancing a full-scale rollout covering 167,000 employees.

UK-based AI chip startup Fractile has raised $220 million (approximately £162.8 million) in a Series B round led by Accel, Factorial Funds, and Founders Fund. Fractile is focused on reducing the time and cost of large-scale AI inference through innovations in chip microarchitecture and manufacturing processes, making high-performance inference more economically viable. The funds will accelerate the delivery of its first chip and systems and expand teams in the UK, US, and Taiwan. The company believes the primary bottleneck in current AI development lies in the cost and efficiency of generating large volumes of output, and its technology aims to overcome this constraint. In 2025, UK tech firms raised over £6 billion in venture capital for AI.

Japan's Liberal Democratic Party (LDP) National Cybersecurity Strategy Headquarters submitted AI and cyberattack countermeasure recommendations to Prime Minister Sanae Takaichi on May 14. The recommendations address risks from malicious use of advanced AI models like Anthropic’s Claude Mythos, calling for fundamental strengthening of cybersecurity measures for highly autonomous AI and enhanced coordination among the Financial Services Agency, the National Center of Incident Readiness and Strategy for Cybersecurity (NISC), and the Artificial Intelligence Strategy Institute (AISI). The report also highlights concerns about foreign actors using AI to generate disinformation, urging stronger information sharing with platform companies. Prime Minister Takaichi stated that the rapid advancement of AI makes this a 'race against time' and emphasized the urgency of swift policy responses.