July 2026: OpenAI Agent Incident, KDDI Breach

July 2026 saw several high-impact security incidents: large credential and personal-data exposures (KDDI: ~12.2M emails and ~7.6M passwords; AssuranceAmerica: ~7M people; MCBS: ~1.26M), an alleged Accenture data dump containing keys and tokens, an AI-agent–driven compromise involving OpenAI and Hugging Face that chained trust-boundary failures and exposed credentials, a widespread GitHub-based infostealer distribution campaign, an extortion attempt on Stadler Rail via a supplier platform, and the disruption of the NetNut residential proxy service — highlighting risks from exposed secrets, third-party platforms, supply-chain paths, and malicious automation.