AI and Oregon Businesses: Risks & Opportunities

This forum summary was generated by AI, based on the forum transcript, and checked for accuracy by OCBA staff.

Event Overview

The Oregon City Business Alliance March Forum tackled one of the most consequential topics facing businesses and communities today: the rapid rise of artificial intelligence. Hosted by OCBA Board Chair Kent Zeigler and introduced by Board Member William Gifford, the forum assembled a panel of five experts representing state government, labor economics, county administration, city technology, and higher education.

The conversation was wide-ranging and honest. Panelists didn’t shy away from the disruptions AI is already causing—job displacement, cybersecurity threats, deepfakes, hallucinations in public records—but they were equally emphatic about its promise: efficiency gains for small businesses, better public services, and a workforce transition that, managed well, doesn’t have to mean mass unemployment.

The forum’s central message was both grounding and urgent: AI isn’t a future event. It is happening right now, in county offices, city departments, and college classrooms—often without a policy framework in place to govern it. The businesses and institutions that adapt thoughtfully, keeping humans at the center of decision-making, will be best positioned to benefit from what comes next.

---

Speakers

William Gifford — Board Member, OCBA (Moderator & Opening Presenter)

William Gifford opened the forum with a global overview of AI, covering international policy frameworks, the competitive landscape between the U.S. and China, and the economic stakes involved. He framed AI as neither inherently good nor bad—“just a tool”—that demands intelligent, intentional adoption.

Senator Mark Meek — Oregon State Senator, District 20

Senator Meek brought a legislative perspective, walking attendees through the state’s AI governance framework, recent bills passed in the Oregon Legislature, and the tension between federal and state approaches to AI regulation. He also highlighted specific opportunities for small and mid-sized businesses in Clackamas County.

Lynn Wallis — Workforce Analyst / Economist, Oregon Employment Department

Lynn Wallis presented data-driven analysis from the Oregon Employment Department and national research bodies including the Pew Research Center and the National Bureau of Economic Research, focusing on which occupations face the highest AI exposure and what that means for the regional workforce.

Andrew Jarocki — Advisor / Clerk to the Board, Clackamas County

Andrew Jarocki offered a candid look at how Clackamas County is currently using over 130 AI tools across its departments, the cybersecurity threats the county faces monthly, and the policy framework it is building to govern AI use responsibly across its 1,000-plus employee workforce.

Michael Dobaj — IT Director, City of Oregon City

Michael Dobaj covered AI from a practical IT and small-business lens, including the different types of AI, common risks like shadow AI and data exposure, and the opportunity for organizations of any size to use AI to amplify their capacity. He also presented highlights on behalf of Kelly Hart, an economic analyst who could not attend.

Dr. Katrina Boone — Associate Dean, Division of Institutional Effectiveness & Planning, Clackamas Community College

Dr. Boone closed the panel with a higher-education perspective, discussing the Generative AI Policy and Practice Task Force she formed at CCC, the questions the academic community is wrestling with around academic integrity and workforce preparation, and her core conviction that human judgment must remain at the center of any AI implementation.

---

By the Numbers

130+	AI tools currently in active use across Clackamas County departments
100+	Cyberattacks experienced by Clackamas County every single month
21%	of U.S. workers reported using AI for at least some of their work in 2025, up from 16% the prior year
50%	of U.S. adults felt more concerned about AI in their daily lives in 2025, up from 37% the year before
77%	of organizations at the 2025 World Economic Forum said they would respond to AI by reskilling existing workers
41%	said they would downsize their workforce where AI could duplicate workers’ roles
86%	of workers with high AI exposure and low adaptive capacity are women
17%	of 849 Oregon occupations projected to see negative job growth through 2034
$2.6–$4.4T	annual value McKinsey estimates generative AI could add across industries
$700B	in capital expenditure planned by five major U.S. tech companies in AI infrastructure
5.6 hrs	saved per worker per week on average through AI tool adoption. (This one is debatable, with some research pointing towards 2.2 hours saved instead)
82%	of small businesses using AI increased their workforce over the past year

---

Key Takeaways

AI Is Already Here—This Is Not a Future Conversation

Multiple panelists drove home the same point from different angles: AI is not something that is coming. It is already embedded in daily operations across county government, city departments, colleges, and businesses. Clackamas County is running over 130 AI tools today. The City of Oregon City uses AI in its cybersecurity stack and its street sweeper fleet. Clackamas Community College’s Microsoft-based tools already have AI built in, with or without a formal policy. For business owners in the room, the question is not whether to engage with AI—it is whether to engage intentionally.

The Workforce Will Change—But Not Simply Disappear

Lynn Wallis presented the nuanced picture that national research offers: AI will most heavily affect clerical, administrative, and sales occupations—and 86% of the workers in those high-exposure, low-adaptability roles are women. Oregon’s own employment projections show 17% of occupations trending toward negative job growth through 2034. But the World Economic Forum’s survey of business leaders found that 77% plan to respond to AI by reskilling workers, and only 41% plan workforce reductions. The conclusion: jobs will change more than they will disappear, and preparation matters enormously.

Small Businesses Have More to Gain Than They Realize

Michael Dobaj, presenting both from his city IT role and from notes prepared by Kelly Hart, made a compelling case for small-business AI adoption. The efficiency gains are real: AI tools save an average of 5.6 hours per worker per week, and 82% of small businesses that adopted AI grew their workforce in the past year. The opportunity is especially significant for organizations with limited staff where every hour of capacity matters. McKinsey estimates that generative AI could add $2.6 to $4.4 trillion in annual value across industries, and much of that flows through the kind of routine tasks—customer service, marketing drafts, administrative summaries—that consume disproportionate time at smaller organizations.

Cybersecurity Is the Most Urgent Dual-Use Problem

Both the county and city representatives gave sobering accounts of AI’s role in cybersecurity. Clackamas County experiences over 100 attacks per month from hackers globally, and AI tools now act as force multipliers for the county’s defense—spotting suspicious login patterns at speeds no human analyst could match. But the same dynamic applies to attackers: AI now enables highly sophisticated, personalized phishing and impersonation attempts. One example cited involved a deepfake audio call where someone impersonated a CEO and successfully prompted a $220,000 wire transfer in the UK. The advice to businesses was direct: establish verification protocols, create family and team “safe words,” and never act on urgent financial requests without independent confirmation.

Hallucinations and AI-Generated Errors Are a Real Governance Problem

Andrew Jarocki highlighted an underappreciated risk: AI-generated content submitted to public bodies that contains fabricated citations, non-existent cases, and invented facts. The county is already receiving materials from constituents that rely on AI tools and suffer from hallucinations. A police department in Utah made headlines after its AI report-writing tool claimed an officer had “transformed into a frog” during a call—because a TV in the background was playing The Princess and the Frog. These aren’t edge cases; they are structural limitations of current AI tools that require human review at every stage where accuracy matters.

See also: What is an AI Hallucination (IBM)

Policy Is Racing to Keep Up—With Mixed Results

Senator Meek walked through Oregon’s governance work in detail: Governor Kotek’s 2023 executive order establishing the AI Advisory Council, its 74 specific recommendations organized around 12 guiding principles, and legislation passed this session requiring disclosure of AI-generated campaign content, protecting professionals from AI impersonation, and adding consumer safeguards around AI chatbots—including mandatory crisis redirection to 988 after troubling reports of AI encouraging self-harm in young users. At the federal level, President Trump’s December 2025 executive order directing the DOJ to challenge state AI laws it deems too burdensome adds complexity. Oregon’s approach, the senator argued, should be practical and targeted—ensuring it works for Oregon businesses and families, not chasing every restriction.

People First—Always

Dr. Boone brought the forum’s most human-centered perspective. Having formed Clackamas Community College’s Generative AI Policy and Practice Task Force, she has spent months in conversation with faculty, staff, and students about what AI means for education, academic integrity, and the future of learning. Her bottom line: “People first. Humans first. Always.” AI must remain a tool that increases human capacity, not one that displaces human judgment, relationships, or decision-making. Her practical takeaway: write down five ways your industry must remain human-first, and five AI skills you commit to learning before you fall behind.

---

Speaker Highlights

William Gifford — Opening Overview

William set the global context with a broad survey of AI’s current landscape. He traced international governance efforts—the UK’s Bletchley Declaration, the Seoul AI Safety Institute, India’s global AI summit—and noted that democratization of AI is a shared international goal, even as competitive dynamics with China (including DeepSeek’s emergence as a low-cost alternative to Western AI platforms) complicate that picture.

He highlighted the economic stakes: five U.S. tech companies are committing $700 billion in AI capital expenditure. Accenture—with $70 billion in revenue—has retrained 550,000 employees on AI, and its CEO made clear that staff who cannot adapt face termination. Anthropic’s research uses “exposure” to describe how vulnerable a job category is to AI automation, and their analysis shows that virtually every occupational category has meaningful exposure.

His closing framing: AI is no more stoppable than electricity or agriculture. The word “sabotage” comes from the French sabot—the wooden shoe workers threw into the looms of the Industrial Revolution. It didn’t stop the looms. The goal now is to learn to use the tools safely and wisely.

Senator Mark Meek — State Policy

Senator Meek grounded his remarks in the legislative reality Oregon businesses are navigating. He noted that federal pressure—through the Trump administration’s executive order tying federal funding to states that reduce AI regulation—creates real tension with Oregon’s preference for targeted, practical oversight.

At the state level, he walked through the AI Advisory Council’s action plan: 74 recommendations, 12 guiding principles, and five executive actions covering governance frameworks, privacy impact assessments, security architecture, and workforce upskilling. He highlighted three legislative wins: required disclosure of AI-generated campaign content; protection of licensed professionals from AI impersonation; and Senate Bill 1546—a bipartisan consumer protection bill requiring chatbots to disclose they are AI, include safety protocols for crisis situations, and add specific safeguards for minors.

On opportunity, he pointed to the $10 million from the Oregon CHIPS Act directed toward AI and semiconductor workforce development, tied to a 2025 MOU with Nvidia. This funding flows through programs at Clackamas Community College and other providers, including the Generative AI Policy and Practice (GAPP) Task Force that Dr. Boone leads.

Lynn Wallis — Labor Market & Workforce Economics

Lynn brought the most data-intensive presentation of the forum, drawing on Pew Research, the National Bureau of Economic Research, Oregon Employment Department projections, and the 2025 World Economic Forum in Davos.

Her framing: workers in clerical and administrative roles face high AI exposure combined with low adaptive capacity—meaning they’re most likely to be displaced and least likely to transition easily. Critically, 86% of this vulnerable group are women. This is not a future concern; it is a present-day equity issue that workforce planning and education must address now.

Oregon’s own projections show office and administrative support and sales-related occupations leading in negative job growth through 2034. The Davos data offered cautious optimism: business leaders overwhelmingly plan to reskill rather than simply downsize. But that requires intentional investment—and the top skills they’re looking for by 2030 include analytical thinking, creative thinking, AI and big data literacy, and resilience.

Andrew Jarocki — Clackamas County

Andrew’s presentation was a model of institutional candor—acknowledging both the remarkable things AI is already doing for the county and the serious risks it is managing.

On the promise: over 130 AI tools are in daily use, from improved GIS mapping to medication dosage prediction at health clinics. The cybersecurity AI acts as a shadow analyst, detecting suspicious login patterns across enormous data volumes instantly. A forthcoming county chatbot will help residents navigate county services faster.

On the peril: the county receives over 100 cyberattacks per month. Deepfake impersonation of county officials has already happened. And because all county business is posted publicly for transparency, attackers can build highly specific, credible impersonation attempts using real contract data and real names. His conclusion: government will move slower than the private sector deliberately—because moving together and getting it right matters more than moving fast.

Michael Dobaj — City of Oregon City / Small Business Perspective

Michael covered practical ground bridging his IT director role with the small-business economic content Kelly Hart prepared. He walked through the taxonomy of AI types and gave concrete local examples: Oregon City’s street sweeper program uses computer vision to detect and report potholes automatically. His team uses AI cybersecurity tools around the clock—something a small team couldn’t otherwise sustain.

His key risk warnings for businesses: be careful what data you feed any AI tool; shadow AI (staff using unapproved tools without IT visibility) is a genuine threat; deepfake audio can be generated from just three seconds of someone’s voice; and over-reliance on AI output without human validation is where real errors happen. He demonstrated with a watch-face example: when asked to generate an image showing 5:00, the AI produced a digital display reading “5:00” but with analog hands at 10:10—because that’s what stock photography taught it a watch looks like.

On opportunity: the forum’s own slide deck was built using Gamma AI in 15 minutes—something that would have taken an hour or two manually. For small businesses with limited staff, AI’s greatest value is reducing the cost of the blank page and freeing people for higher-value work.

Dr. Katrina Boone — Clackamas Community College

Dr. Boone opened with a crucial reframe: we have been here before. Calculators were once feared as cheating tools—then became required equipment. Online learning was dismissed as inferior—then became essential. AI is following the same arc, and the institutions that will navigate it best are those that approach it with critical thinking and pragmatic adaptation.

She described the Generative AI Policy and Practice Task Force she formed at CCC in summer 2025 as a deliberately slow-moving, conversation-first body. Early meetings were not about policy—they were about listening. The range of student perspectives she encountered mirrors what businesses will find among their own teams: enthusiastic early adopters, principled resisters, and a large middle group trying to figure out where the guardrails are.

The Oregon community college network is working on shared AI literacy resources through the distance learning consortium ADELA, with all 17 community colleges contributing to a shared AI literacy course for educators.

Her closing challenge was personal and practical: leave today and write down five ways your work must remain human-first, and five AI skills you commit to developing before you fall behind. “AI is not going to take your job. The person who knows how to use AI effectively might.”

---

Conclusion

The March OCBA Forum landed with a message that was both energizing and uncomfortable: the AI transition is not something businesses can wait on. It is underway—in government offices, on college campuses, inside the tools employees already use every day. The organizations that are getting it right are moving deliberately, not recklessly: learning what tools do, understanding what data risks they carry, keeping humans involved in consequential decisions, and building the literacy to know the difference between a useful capability and a hallucination.

For small and mid-sized businesses in Clackamas County, the opportunity is real. Efficiency gains are measurable. Costs of entry are lower than ever. The workforce data, while sobering, also shows that most business leaders plan to grow through AI—not shrink. The key is intentionality: start with one task, validate the output, and build from there.

And through it all, the forum’s closing note from Dr. Boone is worth carrying forward: people first, humans first, always. The most sophisticated AI tool in the world still needs a human to ask the right question—and another human to check whether the answer is true (And in this case it is, thank you Claude).