ICE Joins Anthropic's Project Glasswing to Bolster AI-Powered Cybersecurity
In an era where financial institutions face increasingly sophisticated cyber threats, Intercontinental Exchange (ICE) is taking a bold step forward. At the start of June, ICE announced its participation in Anthropic's exclusive cybersecurity initiative, Project Glasswing, deploying the advanced Claude Mythos Preview model across its operations — including those of the iconic New York Stock Exchange — to proactively identify and remediate vulnerabilities before malicious actors can exploit them.
This move positions ICE among a select group of organizations at the cutting edge of AI-driven security, and it signals a broader shift in how the financial industry is approaching the protection of critical infrastructure. In an interview with HousingWire, Bob Hart, president of ICE Mortgage Technology, and Steve Pugh, ICE's chief information security officer, offered an inside look at what the initiative involves and why it matters.
What Is Project Glasswing?
Project Glasswing is Anthropic's invitation-only cybersecurity program built around its most advanced AI model to date, Claude Mythos Preview. According to Steve Pugh, the model emerged shortly after the RSA Conference — one of the most prominent cybersecurity events in the world — and immediately generated significant excitement within the industry.
"Mythos came out right after the RSA conference," Pugh explained. "It was [Anthropic's] next generation of model. They had a core group of folks within that program to really help them figure out what to do with this thing. There was a lot of excitement, but they knew it was pretty powerful. So they created this program called Project Glasswing."
Because of the model's significant capabilities, Anthropic chose a careful, controlled rollout rather than a broad public release. Instead of making Claude Mythos Preview widely available, the company assembled a curated group of trusted organizations — including entities within the U.S. government — to deploy the technology in real-world environments and help shape how it is responsibly used at scale.
For further information on the initiative, Anthropic has directed interested parties to https://www.anthropic.com/glasswing.
Why ICE Is an Ideal Candidate for This Initiative
ICE is not a typical technology company. As the parent organization of the New York Stock Exchange and ICE Mortgage Technology, among many other financial market infrastructure businesses, ICE manages systems that underpin trillions of dollars in daily economic activity. A successful cyberattack on any part of this ecosystem would not just affect ICE — it could send shockwaves through global markets.
This makes ICE exactly the kind of organization for which Project Glasswing was designed. The stakes are high, the data is sensitive, and the consequences of a security failure are catastrophic. By bringing Claude Mythos Preview into its security operations, ICE is leveraging state-of-the-art AI not just to react to threats, but to anticipate and neutralize them.
Importantly, ICE stressed that it is overseeing the AI tool's deployment, security architecture, and governance internally. This is not a case of simply plugging in an AI model and hoping for the best — ICE has taken deliberate, structured steps to ensure the technology is integrated responsibly within its operational and compliance frameworks.
How Claude Mythos Preview Strengthens Cybersecurity Resilience
Traditional cybersecurity approaches are largely reactive. Security teams monitor systems, detect anomalies, and respond to incidents after they have already occurred or are in progress. While this model has served the industry for decades, the growing complexity of financial technology infrastructure — combined with the increasing sophistication of threat actors — demands a more proactive posture.
This is where AI-powered tools like Claude Mythos Preview offer a meaningful advantage. By continuously analyzing systems, identifying configuration weaknesses, mapping potential attack surfaces, and flagging vulnerabilities, the model enables security teams to act before an exploit is ever attempted. The goal, as Pugh articulated, is to shift from a defensive crouch to an offensive stance in cybersecurity operations.
For ICE Mortgage Technology specifically, this has significant implications. The mortgage technology ecosystem handles enormous volumes of sensitive consumer financial data, and any breach in that environment carries both regulatory and reputational consequences. Deploying advanced AI to harden those defenses is not just a competitive advantage — it is increasingly a business necessity.
The Broader Significance for Financial Services Cybersecurity
ICE's participation in Project Glasswing is emblematic of a larger trend: the financial services industry is rapidly recognizing AI not just as a tool for efficiency and customer experience, but as a foundational component of cybersecurity strategy.
Several factors are driving this shift:
- Expanding attack surfaces: As financial institutions adopt more cloud infrastructure, third-party integrations, and digital-first workflows, the number of potential entry points for attackers grows substantially. AI tools that can map and monitor these surfaces continuously are invaluable.
- Regulatory pressure: Financial regulators globally are increasing their expectations around cybersecurity governance. Demonstrating the use of sophisticated, proactive security tools can support compliance narratives and risk management disclosures.
- Speed of threats: Cybercriminals are themselves using AI to craft more effective attacks, automate phishing campaigns, and discover vulnerabilities faster than ever before. Defending against AI-powered threats increasingly requires AI-powered defenses.
- Talent constraints: The global cybersecurity talent shortage means that human analysts alone cannot keep pace with the volume and complexity of modern threats. AI tools multiply the effectiveness of existing security teams.
ICE's Internal Governance Framework
One of the most noteworthy aspects of ICE's approach to Project Glasswing is its emphasis on internal governance. Rather than outsourcing oversight entirely, ICE has taken ownership of the deployment architecture, ensuring that security controls, access policies, and data handling practices meet its own high standards — and those of its regulators.
This approach reflects a mature understanding of what responsible AI adoption looks like in a heavily regulated industry. The power of a model like Claude Mythos Preview comes with a corresponding responsibility to deploy it thoughtfully, with clear accountability structures and ongoing human oversight built into the process from day one.
What This Means for the Future of AI in Financial Security
ICE's entry into Project Glasswing may well serve as a template for how other major financial institutions approach AI-powered cybersecurity in the years ahead. The combination of a capable, frontier AI model with robust internal governance, clearly defined use cases, and executive-level sponsorship represents best practice for the industry.
As Bob Hart and Steve Pugh's comments make clear, this is not a speculative experiment. Just a few weeks into the initiative, ICE leadership was already articulating a confident vision for how AI strengthens the company's cybersecurity resilience — not in the abstract, but in operational terms that connect directly to the protection of critical financial infrastructure relied upon by millions of market participants every day.
For an industry where trust is everything, that kind of proactive, transparent approach to cybersecurity investment sends a powerful signal — to regulators, to counterparties, and to the markets themselves.