Fifth Third Joins Anthropic's Project Glasswing Initiative
Fifth Third Bancorp, the ninth-largest U.S. bank by assets, was invited within the past several weeks into Anthropic's Project Glasswing cybersecurity initiative, Seeking Alpha and PYMNTS reported June 10, 2026. Fifth Third's CFO, Bryan Preston, said at the Morgan Stanley U.S. Financials Conference that the invitation reflected the bank's role in the payments ecosystem, including its Direct Express federal benefits card program, U.S. Customs processing, and payroll processing volume. The bank joins a broader expansion Anthropic announced June 2: about 150 new organizations across 15-plus countries, many in power, water, healthcare, and communications, are gaining access to Claude Mythos Preview, which Anthropic says has already surfaced more than 10,000 high- or critical-severity vulnerabilities among its roughly 50 original partners. For practitioners, this is a concrete data point on how frontier AI cyber capabilities are being rationed to critical-infrastructure operators ahead of wider release.
Fifth Third's inclusion is a useful, named data point in a story mostly reported at the program level: what actually gets an individual bank onto Anthropic's list, and it's revealing that the bank's own CFO ties it explicitly to systemic-payments infrastructure, a Treasury benefits program, customs processing, payroll volume, rather than to any AI strategy of Fifth Third's own.
What happened
Fifth Third Bancorp was granted access to Project Glasswing within the past several weeks, Seeking Alpha reported June 10, 2026, a detail PYMNTS confirmed with a quote from Fifth Third CFO Bryan Preston at the Morgan Stanley U.S. Financials Conference. "We think it was a reflection of just the role we play in the payments ecosystem in the country today, whether it's the Direct Express business, some of the processing that we do for U.S. Customs as well as just the magnitude of payroll processing that we do for the country," Preston said. Fifth Third administers Direct Express, the U.S. Treasury's prepaid debit card program for federal benefits, under a five-year agent agreement, and became the ninth-largest U.S. bank by assets after its merger with Comerica closed February 2, 2026.
Industry context
Fifth Third's invitation sits inside a larger expansion Anthropic announced in its own post on June 2, 2026: after roughly 50 initial partners used Claude Mythos Preview since April to scan their codebases and surface more than 10,000 high- or critical-severity vulnerabilities, Anthropic is extending access to about 150 new organizations across more than 15 countries. The new cohort adds sectors underrepresented in the original group, including power, water, healthcare, communications, and hardware, and includes vendors whose codebases underpin many other organizations, including governments; Anthropic says a successful attack on most partners' systems could affect more than 100 million people. On June 9, Anthropic said its newly launched Mythos 5 model would initially be available only through Project Glasswing, in collaboration with the U.S. government.
For practitioners
For security and compliance teams, Project Glasswing is worth tracking less as a partnership announcement and more as an early signal of how a frontier AI lab is gating access to its most capable cyber-offense-adjacent model, by infrastructure criticality and government coordination rather than by commercial deal size. Organizations that see themselves as systemically important to payments, utilities, or communications infrastructure may want to understand Anthropic's stated security-requirement bar for admission.
What to watch
Neither Anthropic's announcement nor the reporting on Fifth Third specifies technical integration details, such as how vulnerability findings are triaged, disclosed, or patched inside a bank the size of Fifth Third. Anthropic has said its broader goal is shifting support from finding vulnerabilities toward disclosing, fixing, and deploying patches, and plans to expand a separate Cyber Verification Program; whether Project Glasswing participants publish concrete before/after security outcomes is worth watching as the program scales.
Key Points
- 1Fifth Third Bancorp, the ninth-largest U.S. bank by assets, was invited into Anthropic's Project Glasswing cybersecurity program within the past several weeks.
- 2Fifth Third's CFO tied the invitation to the bank's Direct Express, U.S. Customs, and payroll-processing role rather than any AI initiative of its own.
- 3The bank joins roughly 150 new organizations Anthropic added on June 2 after Mythos Preview found over 10,000 severe vulnerabilities among initial partners.
Scoring Rationale
A concrete, named example, the ninth-largest U.S. bank, of a critical-infrastructure institution gaining access to Anthropic's frontier cyber-defense model, with a verified CFO quote explaining the selection rationale and confirmed program-level facts (150-organization expansion, 10,000+ vulnerabilities found, Mythos 5 gated through the program with U.S. government involvement). Raised modestly from the stored 5.6 given the richer verification and the program's national-security relevance; capped below 'major' since this is one bank's onboarding within a broader, already-announced initiative.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
