Newsfeed

Artificial Intelligence Talent Shift: OpenAI Co-Founder Andrej Karpathy Joins Anthropic

In a major artificial intelligence talent shift, OpenAI co-founder and former Tesla AI director Andrej Karpathy has officially joined rival AI lab Anthropic. Announced on May 19, 2026, this strategic hire highlights Anthropic’s growing appeal to elite technical minds within the machine learning industry. Karpathy is expected to focus on pre-training research, specifically utilizing the Claude AI model to accelerate recursive self-improvement. This transition underscores the intense competition for top-tier machine learning expertise as tech giants race to develop the next generation of generative AI models.

Sources:

California Autonomous Vehicles: DMV Expands Oversight for Heavy-Duty Testing

The California Department of Motor Vehicles (DMV) has adopted new regulations to expand oversight and enforcement authority over California autonomous vehicles. The updated rules create a clear regulatory pathway for heavy-duty autonomous vehicle testing and deployment by removing prior restrictions on self-driving trucks weighing over 10,001 pounds. Furthermore, the new DMV regulations introduce phased testing requirements, safety case documentation, and minimum testing mileage thresholds before commercial autonomous deployment is permitted on public roads.

Sources:

Cloud Infrastructure: Cloudflare Completes AI Agent Infrastructure Stack

Cloudflare has officially completed its six-layer AI agent infrastructure stack by rebuilding Browser Run on its Containers platform. The updated cloud infrastructure architecture delivers four times higher concurrency and 50% faster response times for developers. This comprehensive technology stack now includes compute, orchestration, memory, browsing, and commerce layers. The latest development addresses the rapidly growing demand for AI agent capacity and firmly positions Cloudflare as a leading managed agent platform provider in the cloud computing space.

Sources:

Cybersecurity Alert: GitHub Breached via Poisoned VS Code Extension

GitHub has confirmed a major cybersecurity breach resulting in the exfiltration of approximately 3,800 internal repositories by a threat group known as TeamPCP. The security compromise occurred after a GitHub employee installed a malicious version of the popular “Nx Console” VS Code extension. Although this poisoned extension was live on the Visual Studio Marketplace for only 18 minutes, it successfully harvested sensitive credentials, including 1Password vaults, GitHub tokens, and SSH keys.

The attackers leveraged these stolen credentials to clone internal GitHub repositories. This incident is part of a broader software supply-chain attack dubbed “Megalodon,” which has compromised over 5,500 repositories across GitHub using automated workflows and disguised system identities. While GitHub stated that no customer data stored outside its internal repositories was affected, the cyberattack underscores the growing vulnerability of developer tools and the critical need for robust endpoint security in software engineering.

Sources: