Daily News, July 27th - AI token underground

We highlight the escalating threat of scans targeting Java Spring Boot's "/actuator/heapdump" endpoint, a significant data leakage vector, and the continued exploitation of weak login credentials in ESAFENET CDG 3 Document Management Systems. These incidents underscore the persistent danger of misconfigurations and basic vulnerabilities, even in security-focused products.We also examine an emerging underground market for reselling LLM tokens, a trend indicating unauthorized AI resource consumption that demands vigilant API key monitoring. On a positive note, we discuss advancements in AI and security automation, including GitHub and PyPI's new time-based defenses in Dependabot, designed to mitigate supply chain attack impacts. Additionally, we point to valuable resources like Google Cloud Tech's guide on building BigQuery AI agents, offering practical avenues for leveraging AI in security. Key takeaways include immediately securing Spring Boot endpoints, conducting thorough audits of document management systems, and evaluating AI agent tooling for internal automation.