Nozomi joins Anthropic's Project Glasswing on OT security

Nozomi Networks has joined Anthropic's Project Glasswing, adding an operational technology and IoT security specialist to the critical infrastructure cyber initiative.

It will use advanced AI models in its platform to discover vulnerabilities in operational technology, Internet of Things and cyber-physical systems. It will also contribute research insights to Anthropic and share findings with the broader cybersecurity community.

Project Glasswing brings together critical infrastructure providers and software groups to use AI for defensive security work. The goal is to identify and mitigate vulnerabilities before attackers can exploit them, with a particular focus on software and systems used in essential services.

Nozomi's entry into the programme reflects growing interest in applying AI to industrial and connected-device environments, where cyber risks differ from those in conventional IT networks. Operators in these settings often manage equipment with long service lives, limited patching opportunities and the risk of physical disruption if systems are compromised.

Critical systems

Those constraints have made operational technology and cyber-physical systems a distinct focus for security vendors and infrastructure owners. Unlike office IT systems, industrial networks in utilities, transport, manufacturing and other sectors often include legacy assets that cannot easily be taken offline for updates or replacement.

Anthropic has positioned Project Glasswing as a collaborative effort between technology suppliers, infrastructure operators and security specialists. The initiative centres on using AI models to improve vulnerability discovery while keeping the work focused on defensive purposes.

For Nozomi, the programme gives it a role in shaping how those methods are applied in operational settings. The company works with critical infrastructure operators across Australia and New Zealand, where industrial cyber resilience has become a priority for asset owners and government agencies.

AI research

Across the industry, vulnerability discovery and assessment are shifting. AI tools are increasingly being tested for security research because they can analyse code, configurations and system relationships faster than manual processes alone. At the same time, those advances have raised concerns that attackers could adopt similar techniques.

Nozomi said AI models such as Mythos are on track to change how vulnerabilities are identified. Within Project Glasswing, it plans to apply those models to the specialised requirements of OT and IoT environments rather than rely on methods designed primarily for enterprise IT systems.

Nozomi has spent more than a decade in the OT, IoT and CPS security market. It has been developing and training its AI engine internally since 2013, using data and observations gathered from thousands of real-world OT and IoT environments.

That background is likely to be central to its contribution to the group, as industrial systems generate different data patterns and operate under different constraints from conventional networks. Security teams in these environments often balance cyber risk against safety, uptime and regulatory obligations, limiting the options available when a flaw is found.

Sector focus

Project Glasswing's focus on critical infrastructure also underscores the sector's strategic importance for AI-led security work. Power, water, transport, manufacturing and other essential services have become a growing focus for cyber defence programmes because incidents can disrupt supply chains, public services and physical operations.

Nozomi said the programme is intended to ensure the requirements of OT and CPS environments are represented in the next stage of AI-driven cybersecurity work. That includes the realities of long asset lifecycles, constrained maintenance windows and the potential for direct physical consequences from cyber incidents.

It will apply advanced AI models to OT and IoT-focused vulnerability discovery in its own platform, contribute insights to Anthropic's collaborative research and share findings with the broader cybersecurity community.