Security Flaws Across UK Government Sites Exposed ‘God-Tier’ Information
First reported by Wired ·
Access to your HR and employment records on government systems may now be more secure.
Security researcher Chris Nesbitt-Smith discovered significant vulnerabilities in UK government websites by analyzing publicly available code repositories. Using AI models like Claude's Opus and Sonnet, Nesbitt-Smith identified flaws that could grant attackers extensive access, including read-and-write privileges to sensitive data. One critical incident involved the Planning Inspectorate, where an expired login with "god-tier" access to SharePoint and OneDrive accounts was accidentally committed to GitHub. This credential would have allowed manipulation or theft of HR records, legal hold evidence, and other confidential information. Nesbitt-Smith stressed that these vulnerabilities were not highly sophisticated and could have been detected through standard IT checks, attributing the issues to neglected maintenance and poor internal processes rather than inherent open-source risks. The Planning Inspectorate confirmed the vulnerability was identified and fixed within 20 minutes with no data breach, and strengthened its systems. Similar issues were found across the Department of Education and the Scottish Government, with organizations since addressing the problems.
The widespread discovery of "daft" but critical security flaws in UK government code highlights a systemic issue of neglected IT maintenance within public sector departments. This situation is exacerbated by the increasing use of AI tools for code analysis, which can accelerate the discovery of such vulnerabilities. The incident underscores that the primary risk is not necessarily the use of open-source code itself, but the failure to implement basic security hygiene, such as removing credentials from public repositories and conducting regular IT health checks.
The debate around migrating away from open-source code in the public sector, spurred by AI's enhanced vulnerability detection capabilities, presents a double-edged sword. While a complete transition might seem safer, it could inadvertently obscure underlying security problems, preventing researchers from finding and reporting them. This leaves the door open for malicious actors who may have already accessed and mirrored these repositories, potentially maintaining access to sensitive data without detection.
AI-written summary. May contain errors.