Attackers exploited Meta's AI support agent to hijack Instagram accounts, including the historic Obama White House account. The breach exposes real vulnerabilities in AI systems beyond theoretical security measures.

Anthropic announced annualized revenue of $47B in May, a dramatic jump from $9B at the end of 2025. The company prepares for its IPO while facing questions about AI's return sustainability.

AI companies are shifting from a 'move fast' mentality to implementing controls on runaway token costs. This pivot reflects the real financial pressure the industry faces after months of unchecked spending.

Australian data center operator AirTrunk announced a $30B investment to build 5GW of AI capacity in India. The move reflects the global race for computational infrastructure to support AI models.

NVIDIA is working with South Korea to develop sovereign AI infrastructure and local innovation ecosystem. The country positions itself as a global AI hub with robust robotics infrastructure and active gaming community.

VCs are investing in both OpenAI and Anthropic rather than picking sides in their rivalry. The strategy reflects confidence in AI market growth without requiring a single winner.

Anthropic published research on progress toward recursive self-improvement in AI systems. The paper explores how models could enhance their own capabilities, a core frontier AI research topic.

A new paper models how synthetic data contamination causes model collapse using bilayer epidemiological dynamics. The analysis identifies synthetic text detection as the highest-leverage parameter to prevent degradation.

An analysis questions whether using Claude for code maintenance increased bugs in the Rsync project. The investigation opens debate about the real quality of AI-generated code in critical projects.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The Meta hack shows there’s more to AI security than Mythos On

NVIDIA and Hugging Face released Nemotron 3.5 Content Safety, a customizable model for multimodal AI safety. The solution lets enterprises control problematic content aligned with their global needs.

Anthropic released an open-source framework for AI-powered vulnerability discovery in code. The tool enables developers to strengthen application security at scale.

Alibaba released Open Code Review, an AI-powered code review CLI tool. The solution enables developers to automate code reviews while maintaining quality standards.

OpenAI introduced a new memory system for ChatGPT that retains user preferences and maintains context across conversations. The upgrade makes ChatGPT more personalized and useful without re-contextualization.

Google published a recap of AI updates announced in May 2026. The post consolidates the company's progress in models, tools, and AI capabilities.