AI Is Critical Infrastructure And It Needs Public-Interest Guardrails

Artificial intelligence and other digital systems have become critical infrastructure, but society has not built the public-interest protections around them that it expects from other essential technologies, says computer scientist and author Francine Berman.

Berman, author of Better Tech: Putting People First in Cyberspace, told Energi Media that search engines, social media, digital finance, education and commerce now depend heavily on technologies that began as private-sector products designed for convenience and profit.

“We don’t apply the same kinds of guardrails and protections to them that we do to other kinds of critical infrastructure,” she said.

Her model for change is the automobile. Cars became indispensable decades before governments, manufacturers and the public developed the safety standards, environmental rules and regulatory institutions that eventually reduced their risks. Berman argues digital technology requires the same long-term approach.

That does not mean stopping innovation. Companies need to make profits and continue investing in research, she said, while governments have a different responsibility: protecting the public. When technologies create serious risks, regulation can establish a common floor for security, privacy and other safeguards.

Berman pointed to the European Union’s AI Act as an important experiment because it links regulatory scrutiny to risk. The higher the potential harm, the stronger the expected guardrails.

AI creates a particular challenge because automated systems increasingly influence or make decisions. Berman cautioned against treating chatbots as though they possess human judgment. “It does not have feelings. It doesn’t have a moral code. It doesn’t have compassion,” she said. For high-risk situations, humans need to remain involved.

The political environment complicates that task. Berman said competition between the United States and China for AI leadership creates pressure to prioritize technological primacy over public protection. She contrasted that approach with the European Union’s stronger regulatory institutions and privacy framework.

The Trump administration’s approach illustrates how quickly the policy environment can change, she said. Different governments can strengthen or dismantle protections, meaning no regulatory settlement will be permanent.

Berman’s prescription is therefore deliberately incremental. There will be no single law or institution capable of fixing the problem. Safer design, regulation, public pressure, education and individual action must reinforce one another over time.

“If each of us do something small, then [the] network effect of people taking action can actually prevail,” she said.

Her broader message is that technological outcomes are not inevitable. Humans made the choices that created today’s digital systems, and humans still have the power to make different ones.