Meta's teen limits and New York's classroom AI ban face Prohibition's old problem

A ban on a tool people already carry covers only the devices and services the rule-maker controls. New York City's classroom rule is built around school-owned hardware ^[7]. Keri Rodrigues, founding president of the National Parents Union, said students are already somewhere else. "They're using AI outside of the education space," she told Business Insider ^[19]. Her prediction for the ban was blunter. "They are going to use all kinds of means to go around the system, but also overconsume. But also overconsume in secret," she said ^[15].

The board-deck version of the case against bans is short. Thirteen years of Prohibition left American drinking near where it started ^[1]. Uttara Ananthakrishnan, a technology professor at the University of Washington, said the data shows flat-out bans on technology people already use rarely work, and that VPNs have made them pretty fruitless ^[18]. "My concern about the bans is that people will go underground," she said ^[12].

That version is weaker than it looks on three counts. Business Insider itself notes that nobody could buy alcohol over a VPN a century ago, and that Prohibition's workarounds were often explicitly criminal ^[17]. The Australian figure comes from Qustodio, a company that sells parental-control tools ^[9]. And the April study by Ananthakrishnan and colleagues is described as showing that users rapidly moved to adult sites without age checks ^[10]. The sentence Business Insider quotes from that study says the laws "did not trigger systematic diversion of users from adopting or exiting platforms toward non-adopting ones" ^[11]. The summary and the quotation point in opposite directions.

Meta's settlement tests coordination. Its limits bind Instagram and Facebook ^[3]. Ananthakrishnan said that without standardized, industry-wide regulation, teens will quickly move to platforms offering unrestricted access ^[13]. "The network is very easy to shift," she said ^[13]. Of social media bans generally, she said: "Enforcing it is incredibly difficult" ^[14].

The sources cover teens, schools and adult websites; none of them studies employees. I'd still expect one part to carry over to a team AI policy. A rule written for company laptops covers the same channel New York's rule covers for school hardware, and it does nothing about a chatbot on a personal phone. New York's high school design is the other model in the record, with two AI literacy courses a year and access in limited educational contexts ^[7]. Rodrigues said she wants policies that teach students to use the tools responsibly and safely ^[16].

The trade-off is control against visibility. A ban on approved systems is quick to write and easy to defend this quarter. If Rodrigues is right about secret use, the cost comes later, as use the policy cannot see or shape. The record is young. Australia's ban dates from last year ^[8], and Meta's settlement and the New York and Los Angeles classroom announcements all came last month ^[3][6].