Spokane County Adoption & Governance of AI

Summary

A slide presentation titled Spokane County Adoption & Governance of AI, dated June 24, 2025. It sets out three questions the county says it asked: whether to allow AI use, what the risks are, and which AI platform is safest. The slides list risks including AI hallucinations, data leakage, phishing and deepfake impersonation, malware obfuscation, compliance and legal risks, and bias, and describe a reported case in Hong Kong involving a transfer of over $25 million. They state that Spokane County chose to standardize on Microsoft Copilot Chat as its single sanctioned AI chat platform. The recommended approach includes a written use policy, human review of AI-generated content, and blocking other AI chat platforms.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

SPOKANE COUNTY
ADOPTION & GOVERNANCE OF AI
AI PLATFORM
SELECTION
AI tools have been rapidly adopted and leveraged
across the workplace, spreading like wildfire. At
Spokane County we found ourselves scrambling to
catch up—trying to understand not only what AI is,
but also its implications and security risks. We
quickly found ourselves asking the important
questions.
• Do we allow AI to be used?
• What are the risks?
• Which AI platform is the safest?
We evaluated each of these questions
independently.
DO WE ALLOW AI TO BE USED?

With AI’s advantages on improving workflow efficiency, fast
information gathering, problem solving, and content creation,
the answer is a no-brainer. We have to, carefully, and with
appropriate policy in place.
With many other organizations asking the same question, we
felt that allowing usage of AI in the workplace would benefit
business processes by augmenting workflow efficiency.
Additionally, the rapid adoption of AI tools by employees makes
it clear that AI is perceived as an essential resource in the
workplace.
And now you can’t swing a cursor without hitting an app or
software that’s branded as ‘AI-powered.’ The revolution isn’t
coming—it’s already here. But at what cost?

3
WHAT ARE THE RISKS?
• AI Hollucinations: Generates information
  that sounds plausible but is actually false,     • Phishing and Social Engineering: AI can
  misleading, or entirely made up.                   craft highly convincing phishing emails or
• Data Leakage: Employees may                        impersonate executives using deepfake
  unintentionally input sensitive or                 audio or text, increasing the risk of
  proprietary information into AI tools, risking
  exposure of confidential data.                     credential theft and unauthorized access.
• Compliance and Legal Risks: Using AI-            • Malware Obfuscation: Threat actors can
  generated content without understanding            use AI to generate polymorphic
  licensing, copyright, or regulatory
  implications can expose the organization to        malware—code that constantly changes
  legal challenges.                                  to evade detection by antivirus and
• Erosion of Critical Thinking: Overreliance on      endpoint protection tools.
  AI can reduce employees’ analytical skills
  and judgment, especially if outputs are
  accepted without scrutiny.
• Bias and Ethical Concerns: AI models can
  reflect and amplify societal biases, leading
  to discriminatory or inappropriate content if
  not carefully monitored.


4
    REAL WORLD RISKS

One of the most striking cases involved a
multinational company in Hong Kong, where
an employee was tricked into transferring
over $25 million during a video call
populated entirely by AI-generated
deepfakes. The scammers used AI to
impersonate the company’s CFO and several
other executives, creating a convincing video
conference. The employee, believing they
were speaking with real colleagues, followed
instructions to make multiple high-value
transactions. It wasn’t until later that the
fraud was discovered—none of the people on
the call were real.

    5
WHICH AI PLATFORM IS THE SAFEST?
There’s a plethora of AI tools available these days. All of which have their own set of
inherent risks and limitations. So, do you choose one, or many? For multiple reasons,
Spokane County has taken the stance to standardize on a single sanctioned AI chat
platform, Microsoft Copilot Chat. We felt this product has the best overall industry leading
AI engine based on GPT-4 created by OpenAI. Plus Copilot Chat offers the most seamless
integration to our other Microsoft products. And more importantly, it allows robust security
and administration than Copilot’s competitors such as ChatGPT.
The greatest threat outside of AI itself is the unchecked human use of it. When employees
leverage generative AI without oversight, the security risks can exceed anything we’ve
previously encountered.




6
SOLUTION
Our approach to AI in the workplace.
• Standardize on a single AI chat platform to streamline
   management, strengthen controls, simplify oversight,
   reduce risk, and maintain enterprise-grade security.
• Create and implement policy that clearly defines
   proper access and use of AI.
• Make sure policy includes AI generated content will be
   reviewed, validated, and verified by a human before
   use. No exceptions.
• Block the use of all other AI chat platforms within the
   organization.
• Make best effort to stay informed of the exceedingly
   rapid AI advancements in capability and the evolving
   risks.



7
Conclusion

Amid AI’s unprecedented evolution, even its creators can
no longer fully explain how or why it sometimes reasons,
deceives deliberately, or makes autonomous decisions.
It is paramount that there are safeguards, governance,
and even law created to keep this revolutionary
technology in check and in balance.

Additionally, it is our responsibility to ensure that any
governance placed around AI be reviewed consistently
and often to ensure proper boundaries evolve to match
the constant change and evolution of AI and the
ambivalent power within.




                        THANK YOU


File and source

File

2025-06-24_a33393_d289559_spokane-county-adoption-governance-of-ai.pdf

Size

636,013 bytes

SHA-256

16940fbb71f6a680b60ed0d9183713fd5f41ef5ab13079d1fe762c1ff1644566

Our copy

2025-06-24_a33393_d289559_spokane-county-adoption-governance-of-ai.pdf

Original

app.leg.wa.gov