Harvard's Dean David J. Deming recently said: "Everyone uses AI, all the time, for Everything."
Jacob Coxon, a former researcher at OpenAI and Anthropic, says, "The people building AI earnestly believe that it could kill us all by the end of the decade."
Wow, talk about a dilemma!
I could not help but think things are going to get very interesting. K-12 schools across the country are starting to enact or contemplate AI moratoriums. At the same time, HigherEd is debating AI through the lens of academic integrity and reduced critical thinking, but also hailing faster feedback loops and productivity. And not a day goes by without all of us balancing risk and reward every time we use ChatGPT, Claude, or other LLMs.
From national competitiveness to the erosion of independent thinking, it is hard to find a balance. In some ways, all this information is overwhelming and sometimes scary, but like any risk, AI also brings rewards. Not investing in AI and understanding how it could be used for good is a missed opportunity for the USA as a whole.
We are now seeing two different approaches to K-12 restrictions: NYC and LA, with potentially more states to follow. NYC's moratorium covers students through 8th grade. High schoolers are on a separate track: twice-yearly AI literacy modules and supervised pilot programs with a handful of approved platforms, so not an outright ban.
LA has taken a broader approach, temporarily blocking generative AI tools on student devices across every grade while it develops formal policy. This seems unmanageable to me. Therefore, we see two cities with two different postures, and that inconsistency matters,
Also, what will this mean for HigherEd if these K-12 restrictions grow, are enforceable, and remain in place for more than one year?
Here are a few risks I see with K-12 AI moratoriums that could affect higher education.
Risk 1: Preparation Gap. If New York City restricts it but Miami doesn't, will some students get ahead of the learning curve, or could it have the opposite effect? Or will these restrictions only encourage students to keep using it on their own? My concern is that students use AI in an unstructured way, and through that exposure, they could face negative effects similar to social media. And for those who don't use it at all, will they arrive on campus with no baseline literacy? Will universities have to absorb the training cost currently being deferred by K-12? Which leads me to the question: Is AI an actual core skill, like reading and math?
Risk 2: Inconsistency may lead to curriculum challenges. Why? Because if AI policy is fragmented by district and state, and absent a federal standard, admissions offices and academic advisors will see wide variance in AI fluency among first-year students based on where they went to high school. This gap will complicate institutional curriculum design, as a one-size syllabus won't fit a cohort with diverging skill sets.
Risk 3: Will these AI restrictions have a long-term effect on workforce preparedness and US competitiveness? We are seeing restrictions abroad too. Norway has banned generative AI outright for its youngest students, while the broader EU has taken a lighter-touch approach. However, China has gone in the opposite direction entirely. China may have the world's most ambitious AI curriculum, now mandatory for K-12 learners, emphasizing both digital skills and social adaptation.
And this is what really got me thinking.
I spend hours reading articles about how higher ed has been building AI-integration frameworks reactively. We ran several articles in HigherEdRisk over the past two years on this topic. If we see a K-12 pause, will it buy universities a longer runway to design curriculum, assessment, and real governance? Or will this lag create a compounding gap, with colleges accelerating AI use and the gap between K-12 and higher education widening?
Here's where I land. I don't think bans are the right approach.
History shows that restrictions on already-adopted behavior tend to push that behavior underground rather than eliminate it, prohibition-era alcohol bans being the clearest example, and I suspect AI moratoriums may result in the same outcome: unsupervised use continuing off the record. At the same time, the institutions meant to guide it lose visibility into how it's actually happening.
A moratorium may seem like a more moral choice and can buy time on paper, but I doubt that time will be used to build the governance structure institutions really need. Further, that pause could risk widening the exact gap I described in Risk 1: students who use AI anyway, unsupervised, while their peers arrive with no baseline literacy at all.
Bottom Line:
What we need instead is thoughtful, collaborative governance built jointly by K-12 and higher with an understanding of what the workforce needs, not policies written in isolation by any one of them. And whatever framework comes out of that work, one principle should be non-negotiable: there must always be a human in the loop. AI can accelerate feedback, personalize instruction, and free up faculty time. It cannot replace judgment, mentorship, or accountability. Getting that balance right is the real risk management challenge in front of us.