US bishops press Capitol Hill on the child safety bill against AI, and Congress leaves without voting on it

By Zendoric · September 18, 2026.

More than a dozen U.S. bishops met on September 16 with senators and representatives at the Capitol to call for effective AI «guardrails» and to press for passage of the Kids Online Safety Act (KOSA, S.1748), the federal bill that seeks to protect minors online. As Archbishop Alexander K. Sample of Portland, Oregon, chairman of the religious liberty committee of the United States Conference of Catholic Bishops (USCCB), told OSV News, the visits took place on the same day the USCCB announced a new working group to engage with technology leaders and lawmakers on AI, human dignity and children's online safety.

The meeting came on a particularly crowded day in Washington: Congress was preparing to adjourn with no expectation of returning until after the November elections, as the article itself notes. KOSA is dogged by a specific disagreement between the two chambers: the House of Representatives passed a version of the bill earlier this year, but the Senate's includes a «duty of care» clause that is absent from the House text, and over which the White House has negotiated with senators. That clause, Sample explained, would place on developers the responsibility for ensuring that protections for minors exist.

The backdrop is the dispute, now firmly lodged in the Capitol, over how to respond to warnings from several technology leaders who have called for slowing AI development in order to set safeguards. President Donald Trump has broadly dismissed that concern and maintains a position in favor of accelerating development, on the argument —according to the article— that slowing down would leave the United States behind China. Some Republicans back that line of industry self-regulation; some Democrats are calling for regulation. Pope Leo XIV himself had called earlier this year, in his encyclical «Magnifica Humanitas», for «adequate regulatory tools capable of upholding justice and curbing the distorting effects of technological power».

Our reading: what matters here is not that the Church has views on AI —it has been voicing them for months, encyclical included— but the timing gap this episode exposes. Technology experts themselves, as Sample recounted, warned them that very morning that «the next 12 to 18 months are going to see tremendous advances». And Congress, in practice, is going on recess without having settled even a narrow bill with relatively broad consensus such as KOSA, focused on minors and not on frontier AI. If specific, already mature legislation moves at this pace, any more ambitious regulatory framework for frontier models is, in effect, postponed indefinitely.

The other point that deserves attention is where the regulatory battle is really being fought: not over whether there should be rules, but over who bears the responsibility. The Senate's «duty of care» is, in essence, the same debate we already see in other jurisdictions —in the European Union, the AI Act is beginning to build a liability and incident regime that falls on whoever deploys the system—: shifting the cost of risk from the user (or their parents) to whoever designs and markets the technology. It is a governance mechanism, not a ban, and that is why it generates less political friction than halting model development. It also explains why the White House is negotiating that specific point: that is where the real compliance cost for companies lies.

The «race with China» argument the Trump administration uses to resist regulation —a theme we have followed in our comparison of the two countries' capabilities— surfaces again here, but this time applied to terrain where it does not compete: protecting minors from addictive content and design does not slow foundational-model R&D against Beijing. Conflating the two debates, as part of the current political discourse does, serves more to block any legislation than to defend a real technological advantage.

The bishops, for their part, deliberately choose not to speak as technicians: Sample was explicit that they address lawmakers «not as experts, but as pastors», focused on the dignity of the person and —more briefly, but present at the meeting— on AI's impact on workers. It is a strategy consistent with the thesis we hold at Zendoric: the underlying problem is not whether AI advances —it will, and in the long run that capability may translate into health, prosperity and time freed up for people— but whether we build the minimum safeguards in time so that the transition does not leave the most vulnerable, starting with minors, paying the price of an experiment without a safety net. That this message comes from a religious institution and not from a technical regulator says as much about the current vacuum as it does about the urgency of the problem itself.