by Larry Magid
In late June, leaders from governments, tech companies, nonprofits, and academia gathered at the United Nation’s Internet Governance Forum in Lillestrøm Norway to talk about a broad range of internet related issues, including risks of artificial intelligence—especially the rapid rise of generative AI, which can create everything from art and code to deepfakes and misinformation. I was there representing ConnectSafely.
In a session on AI governance, Jhalak Kakkar, Executive Director, Centre for Communication Governance, New Delhi, argued that “Regulation and governance is not a bad word,” he added, “We’ve talked about risk, now let’s focus on innovation.” But he also said, “we need to be carrying out AI impact assessments, from a socio-technical perspective so that we really understand impacts on society and individuals.”
Mlindi Mashologu – Deputy Director General, Ministry of Communications and Digital Technology, South Africa put the issue into context, “We are trying to make sure our approach is grounded from the principle of data justice, which puts human rights, economic equity, and environmental sustainability at the center of AI.” And he certainly doesn’t want AI making final decision, “Humans must guide, and when needed, override automated systems. This includes reinforcement learning with human feedback and clear thresholds for interventions in high-risk domains.”
And, as Meta’s director of Privacy Policy, Melinda Claybaugh, pointed out “We already have laws that pre-date ChatGPT—around copyright, data use, misinformation, safety. We should look at how those legal frameworks are fit for purpose or not, rather than rushing to regulate the technology itself.”
Why regulate?
Still, there was a general consensus that AI shouldn’t be left entirely unregulated. AI is reshaping our lives, including how we learn, do research, communicate and even govern. But it comes with some very real risks, including:
- Misinformation: Generative AI can create fake images, videos, and audio that look and sound real and give seemingly logical yet erroneous answers to questions.
- Bias: AI systems can reflect and amplify societal biases in ways that are hard to spot but can cause real harm.
- Child safety: AI is already being misused to create sexually suggestive or explicit deepfake images of minors and there are other issues, including privacy and appropriate use in education.
- Inequality and concentration of power, often concentrated in a few tech companies or countries, potentially exacerbating global inequalities and creating dependencies
I also worry — at least a little — about its impact on cognitive thinking and creativity. I’m happy to use it as a tool to help people create content, but I worry about it diminishing our ability to think for ourselves and authentically express ourselves. People are already using it to write articles and academic papers. While it’s a legitimate to use AI to help the creative process, it shouldn’t replace it.
Avoiding Micromanaging
Trying to regulate AI at the level of code or algorithms isn’t practical—and it could do more harm than good. Technology evolves too quickly for rigid, technical rules to keep up. If governments had tried to freeze the internet’s early protocols, the web as we know it wouldn’t exist. It’s hard enough to craft sensible regulations at the state or national level and even more challenging when it comes to international rules, such as those that are sometimes talked about at UN events like the IGF.
Instead of prescriptive mandates, we need principles-based regulation—broad guardrails that promote safety and accountability without blocking progress. Companies should be free to innovate, as long as they abide by sensible rules that avoid harm. What we mostly need is thoughtful design — not just by the companies themselves — but in collaboration with other stakeholders who will be impacted by their designs.
Smart, Sensible Guardrails
Some common-sense protections are already emerging as part of the global conversation:
- Transparency: People should know when they’re interacting with AI, whether it’s in customer service, content creation, or media.
- Accountability: Companies need to take responsibility when their AI products cause harm—whether that’s bias, safety concerns, or misuse.
- Protecting children: AI must never be used to exploit or endanger young people. That includes addressing the growing problem of AI-generated child sexual abuse material and deepfake harassment.
-
Data Privacy and Consent with data minimization and anonymization and strong data security to prevent misuse or leaks.
Global Cooperation, Not One-Size-Fits-All Rules
One of the clear messages from IGF is that AI governance requires global collaboration. No single country or company can—or should—control the conversation. And, like most IGF discussions, it’s not just about governments. The conversation needs to include tech leaders, researchers, civil society, and everyday users, including youth.
At ConnectSafely, we’ve long said that protecting people—especially children—requires balancing safety and rights. We need systems that prevent harm without undermining free speech, privacy, or access to technology.