BT bolsters network security with Anthropic Claude Mythos AI

BT has joined Anthropic Project Glasswing to deploy the Claude Mythos Preview AI model against network security vulnerabilities.

Telecom operators face continuous pressure to automate threat detection. BT alone prevents four million cyberattacks across its networks every day.

Human operators cannot manually parse the telemetry generated by this volume of hostile traffic. The integration of large language models into security operations centres provides a mechanism to process incident logs faster than traditional heuristic systems.

BT Chief Executive Allison Kirkby confirmed the company’s membership in Anthropic’s Project Glasswing at the UK Government’s AI Adoption Summit. This agreement grants BT access to Claude Mythos Preview.

“AI only works at scale when it is underpinned by future-ready networks that are secure, resilient, safe,” said Kirkby.

Anthropic designates Claude Mythos Preview as a frontier AI model. Project Glasswing brings together essential infrastructure providers to secure data and systems which underpin services that millions of people rely on. Trusted organisations can use Anthropic’s systems to rapidly identify vulnerabilities. Security teams use the model to find flaws and fix them before criminals can take advantage.

BT runs this capability alongside its existing defensive infrastructure. Machine learning algorithms currently identify anomalous traffic patterns across the global network. Generating actionable intelligence from these anomalies remains an intensive task. Claude Mythos Preview addresses this by ingesting raw security alerts, correlating events across different network nodes, and outputting natural language summaries for incident responders.

Jon James, CEO of BT Business, commented: “AI is changing cybersecurity fast, and businesses need trusted partners who can help them stay one step ahead. By joining Project Glasswing, BT will strengthen its own cyber security capability to protect our networks, our customers, and the wider UK.”

Machine speed threat response

Security architectures fail when mean-time-to-respond lags behind adversary execution speed. BT Business recently announced a collaboration with Accenture to develop advanced AI-powered cyber operations. The goal of that partnership involves responding to cyber threats at machine speed. Adding Claude Mythos Preview to this operational stack provides the cognitive engine required to achieve that velocity.

Traditional Security Information and Event Management platforms generate thousands of low-fidelity alerts daily. Human operators dedicate hours distinguishing false positives from actual breaches. The Anthropic model parses these alerts instantly. It reads the code associated with a potential vulnerability, compares the suspicious traffic against known attack vectors, and then delivers a definitive assessment to the human operator.

This design reduces the triage phase from hours to seconds. Network defenders allocate their time executing remediation strategies rather than hunting for the root cause. Fixing vulnerabilities before criminal exploitation forms the exact premise of Project Glasswing.

Building sovereign AI capabilities

Government policy increasingly shapes how telecom providers adopt new technologies. Kirkby opened the UK Government’s AI Adoption Summit in front of senior officials, including Chancellor of the Exchequer Rachel Reeves and Business Secretary Peter Kyle.

Other attendees included DSIT Secretary Liz Kendall, Cabinet Secretary Antonia Romeo, and Minister for AI and Online Safety Kanishka Narayan. The presence of these ministers highlights the intersection of national security and private network operations.

Kirkby emphasised BT’s commitment to working with the government “to support the further development and deployment of sovereign British AI capability.” She articulated a vision where the UK acts as an “AI maker and not just a taker” and positioned BT as an “enabler of responsible adoption and a responsible adopter ourselves.”

Operating as a responsible adopter means isolating the AI model from raw customer data while still allowing it to analyse network telemetry. Telecom providers must maintain strict data governance. Allowing external models to ingest proprietary logs requires secure enclaves.

Anthropic provides enterprise-grade boundaries to prevent data leakage. BT Business already provides AI-powered cybersecurity solutions to customers of all sizes, including new products for small businesses. Protecting the integrity of these commercial offerings relies entirely on the underlying security of the BT network itself.

Methods developed to secure the telecom infrastructure eventually filter down to the commercial products sold to corporate clients. The scale of the threat underlines the importance of staying ahead of it.

See also: Nokia turns 2000km production fibre route into active sensor

Want to learn more about cybersecurity from industry leaders? Check out Cyber Security & Cloud Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the AI & Big Data Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.