AI safety must be the priority

Officials must act now to make safety as important as innovation.

A poll released last week found that most people around the globe are increasingly pessimistic about the state of the world—and that was before the recent flurry of warnings that artificial intelligence is on a path to destroy humanity. Indeed, we’re in an era of massive disruption caused by wars, the deepening affordability crisis, assaults on democracy, and the seismic shifts that AI is triggering in virtually every aspect of society. Yet I am hopeful—perhaps, even, a “hopeaholic,” to borrow a term the late feminist leader Gloria Steinem used to describe herself. My hope stems not from passive wishing but from actively fighting for progress, along with so many others compelled to act by anxiety over the affordability, democracy and AI crises.

Apocalyptic warnings that rogue AI agents could dominate and even extinguish humanity are no longer the stuff of science fiction. President Donald Trump is virtually alone in his assessment that the threats posed by AI are a “HOAX” and that “the only controls or guardrails that AI needs is a strong and smart, high-IQ president.” From political rivals to Pope Leo, there is a growing consensus that, while AI can be a powerful force for good, unchecked AI carries potentially catastrophic consequences and must have guardrails controlled by humans—and the time to act is now. 

Anthropic CEO Dario Amodei’s call for a global slowdown and third-party audits of AI was endorsed even by many of Anthropic’s competitors. Progressive leader Sen. Bernie Sanders and Steve Bannon, the architect of the MAGA movement, put aside their differences last week to call for checks on the potential destructive power of emerging technologies. 

I participated in the event where Sanders and Bannon spoke, the Pro-Human Assembly. I was heartened that so many disparate people—elected officials from both sides of the aisle, labor leaders, parents, artists and leaders of many different faiths—were united in demanding elected officials act to make safety as important as innovation. It’s unconscionable that the federal government hasn’t enacted meaningful safeguards and regulations for AI, but the AFT is taking the lead where we can. 

Building on our “devices down, eyes up, hands-on” plan to boost learning and safety in the AI era, the AFT and the United Federation of Teachers this month reached an agreement with Microsoft for a first-of-its-kind National AI Safety and Privacy Standard for Schools, establishing ironclad data security protections for students, families and educators. While we want legislation broadly enacting such standards, this is a legally binding contract that prioritizes privacy, safety and transparency. This standard ensures student and educator data is private—period. It reinforces that any use of AI requires human oversight and gives families clear information about how AI is being used in schools. Microsoft has agreed that, effective Nov. 1, these standards will be part of all their contracts with school districts.

The AFT is working to get the balance right to harness the benefits of technology while mitigating the harms. That’s why we created the National Academy for AI Instruction last year, to help teachers master AI so AI doesn’t master them. It’s a training hub designed and run by educators, grounded in trust, safety and people-first technology. It builds on the work our members initiated, after ChatGPT was first announced, to develop and continuously update the AFT’s “Commonsense Guardrails for Using Advanced Technology in Schools.” 

In the 37 states that have banned phones during the school day, kids are noticeably more engaged, and hallways and lunchrooms bustle with chatter and laughter again now that students aren’t heads-down, eyes on their phones. One year into its bell-to-bell cellphone ban, Dallas schools are seeing a 24 percent increase in library book checkouts. Imagine the effect on literacy rates with less screen time, more active learning, and kids reading whole books again. 

AI and other emerging technologies can seem so powerful, so unstoppable, that it can feel like there’s nothing we can do to protect ourselves. Amid the many dire warnings of late, I’ve thought of another icon we lost recently, Dolly Parton, who liked to say, “We cannot direct the wind, but we can adjust the sails.” 

The winds sweeping AI over every part of our lives can be turbulent and terrifying. We have no choice but to adjust and guard against the harms. But individuals and even the tech industry cannot manage the AI juggernaut on our own. The American people want leaders to make safety as important as innovation; they want the federal government to take concrete steps now to protect our security and privacy, workers, the climate and humanity itself. That is why the upcoming elections may be the most consequential in my lifetime. We must elect leaders in November and beyond who will act boldly to address affordability, AI and anxiety—before it’s too late.