As regulatory expectations continue to evolve and AI becomes embedded throughout the vendor ecosystem, organizations must rethink how contracts support effective third-party risk management. Elizabeth Blosh-Myers discusses actionable approaches to cyber accountability, resilience requirements, audit rights, AI governance, continuous monitoring, and emerging risks that will shape contract frameworks through 2030 and beyond.
Third-party
relationships are becoming increasingly complex as organizations navigate cyber
threats, heightened regulatory scrutiny, AI adoption, and operational
resilience challenges. In this Q&A, Elizabeth Blosh-Myers, VP Third-Party
and Model Risk Management at Thread Bank, shares practical perspectives on
contract design, vendor accountability, emerging AI risks, and the future
evolution of third-party contracting frameworks.
1. What contractual provisions are most
effective for holding third parties accountable for cyber incidents without
creating obligations vendors will not accept?
Vendors will not agree to
unlimited liability. They will agree to clear rules.
• Tell me within 48 hours if you think
something happened, 24 hours if you know it did. Define "know" or
their investigation becomes my clock.
• Don't fight over the whole liability cap.
Set a higher cap just for breaches and fines. That gets signed.
• Make your subcontractors follow the same
security rules, and let me say no to new ones. Most of the breaches I see come
from the vendor's vendor.
• Agree now to share investigation results,
cooperate with my regulator, and cover notification costs. Costs them nothing
until something goes wrong.
A clause they won't sign
protects no one. A clear trigger and a fix-it period is what I can actually
use.
2. How have contract requirements evolved over
the last 12 to 18 months around resilience, audit rights, incident
notification, and ongoing monitoring?
Examiners used to ask if
I had a contract. Now they ask if the contract lets me prove I'm watching the
vendor.
• Audit rights are their own section now.
What I get, how often, and regulator access if asked.
• Resilience got specific: recovery times,
proof they tested their backup plan this year, and a real exit plan with my
data returned in a usable form. After Synapse, everyone in BaaS knows what
happens without one.
• Monitoring is in the contract: they send
me risk metrics, complaints, and notice if they change owners, move work
offshore, or swap key subcontractors.
• Tennessee law and courts. Not negotiable.
If the contract can't
produce the evidence, the examiner assumes the control isn't there.
3. What are the biggest gaps in current
contracts around AI governance, data usage, model transparency, and liability
for AI-driven outcomes?
Vendors are selling AI
and contracting like it's regular software.
• Don't train your models on our data. Not
even "anonymized" data. That's where the loophole lives.
• Bank model risk rules apply to their
models too, and most contracts give me nothing to validate with. I need
documentation and notice when the model changes.
• Vendors add AI to existing products
mid-contract without telling anyone. Define what counts as AI and require
notice before it touches our data or customers.
• Vendors disclaim responsibility for what
the AI gets wrong. If it affects fraud, lending, or disputes, the regulatory
risk lands on the bank. I need quality commitments, the right to have a human
review, and coverage if their model causes a violation.
The bank owns the outcome
no matter who owns the model. Write the contract that way.
4. Will contracts evolve from static documents
into dynamic risk tools using continuous monitoring, automated reporting, or
real-time triggers?
Yes. The technology is
here. The contract language is what's behind.