NHS pulls open-source code from GitHub amid AI hacking fears, sparking transparency backlash

NHS Orders Removal of Open-Source Code Amid AI Security Concerns

NHS England has instructed staff to withdraw all existing and future open-source code from public platforms like GitHub by May 11, marking a significant departure from the organization's longstanding NHS open-source policy 1

1

. The decision to close-source GitHub repos stems from cybersecurity concerns related to advanced AI models, particularly Anthropic's Mythos, which has been reported as capable of discovering vulnerabilities in virtually any software 2

2

. According to internal guidance seen by The Register, the healthcare giant believes that public repositories "materially increase the risk of unintended disclosure of source code, architectural decisions, configuration detail, and contextual information that may be exploited" 2

2

.

Growing Backlash Against Hiding Source Code Decision

The move has triggered substantial backlash from technology experts, digital rights advocates, and even political figures. An open letter calling on NHS England to reverse its decision has attracted 682 signatures, including digital rights campaigner Cory Doctorow and former UK health secretary Matt Hancock 1

1

. In a LinkedIn post, Hancock stated: "One of the smartest things the NHS has done in recent years is open-source its code. Taxpayers paid for it, so taxpayers should benefit from it. But the practical case is just as strong: open source code is more rigorously tested, more secure, and allows the best minds anywhere in the world to build on top of it" 1

1

.

Contradicting NHS Service Standards and Undermining Transparency

The decision directly contradicts the NHS service standard, which requires staff to make software produced with public money open-source so tools can be built upon, improved, and used without duplicated effort 1

1

. The organization's service manual explicitly states that "public services are built with public money" and unless there's a good reason, code should be made available for reuse 2

2

. NHS sources indicated that very few of the hundreds of repositories contain anything remotely sensitive, with examples including documentation, architecture diagrams, and codebases for internal tools such as web apps for managing clinic times 2

2

.

Experts Question Effectiveness of AI Hacking Risk Response

Security experts argue that withdrawing code from public view will do nothing to improve AI security or address the AI hacking risk posed by computer-hacking AI models 1

1

. Former head of open technology at NHSX, Terence Eden, stated that the code "was all ingested for 'training purposes' years ago" and that "closing now doesn't meaningfully protect you" 2

2

. He emphasized that many serious vulnerabilities facing organizations exist not in their codebases but in their software supply chains 2

2

.

Mythos AI Model Capabilities and Testing Results

Vlad-Stefan Harbuz at the University of Edinburgh, a co-author of the open letter, has access to Mythos and was part of a group that recently used it to scan open-source NHS code for vulnerabilities 1

1

. They found "a few relatively severe vulnerabilities" that were responsibly disclosed to the NHS prior to the decision 1

1

. While national authorities including the UK's AI Safety Institute and National Cyber Security Centre have validated Anthropic's claims of Mythos representing an advancement beyond forecasted AI development cycles, skepticism remains about its purported bug-hunting power, particularly as Anthropic has not revealed the number of false positives the model generates 2

2

.

Temporary Measure with Uncertain Timeline

An NHS England spokesperson characterized this as a temporary measure enacted while the organization strengthens its cybersecurity posture, stating they "will continue to publish source code where there is a clear need" 2

2

. However, NHS England did not provide an estimate for when this temporary closed-sourcing will end or specify what it deems the most significant threats advanced AI models pose to its repositories 2

2

. The decision, approved by the NHS Engineering Board, requires GitHub repositories to be set from public to private unless there is "an explicit and exceptional need" 2

2

. This shift raises questions about the future of transparency in government-funded technology projects and whether the perceived security benefits outweigh the loss of collaborative development and public accountability that open-source code provides.