A former Cornell student, known publicly as Jane Doe, was drugged and raped by seven men at a fraternity house in October 2024. Campus police referred her case to the district attorney. No criminal charges followed. By one account, the DA dismissed the case without ever speaking to her. Her story went viral this month, part of a wave of survivors speaking out, and suddenly the country is paying attention to something that has been true, and ignored, for decades.

The attention is finally forcing into view that this is not one broken case. It's the system working exactly as it has been built to work. American universities suspend roughly 1 in 12,400 enrolled students per year for reported sexual misconduct, and expel about 1 in 22,900. Nearly 40% of student survivors end up taking leave, transferring, or dropping out, & often, the person harmed is the one who leaves. And in 2025, the federal Office for Civil Rights reached zero resolution agreements on major campus sexual-violence cases, having redirected most of its Title IX enforcement toward transgender athletics instead.

Thousands of cases. Quietly unraveling. For years.

And I'd argue this didn't get normalized in a vacuum. It's hard to tell survivors their allegations will be taken seriously when accountability so visibly evaporates the higher up you go. We have a president whom a jury found liable, in a 2023 civil trial, for sexual abuse and defamation. That is a matter of court record, not opinion. When that's the ceiling, the floor gives way. The message travels downward, to every DA who declines to make a call, every Title IX office that finds a reason not to act, and every institution that decides the reputational risk of accountability outweighs the human cost of inaction.

So yes, it's about damn time. The virality is doing what the institutions wouldn't: refusing to let these cases disappear quietly. Because when self-governing breaks all trust, we, the people, must stand up.

And speaking of institutions self-governing behind closed doors and deciding how accountable they're willing to be, this week, the federal government formally took the wheel on who sets the rules for the most powerful technology on earth today.

On October 4th, DJT announced the formation of a federal AI task force he's calling the "Super Intelligence Force" and named Director of National Intelligence, Jay Clayton, to lead it as the administration's new AI czar. The name follows an executive order Trump signed directing federal agencies to use "super intelligence" (SI) instead of "artificial intelligence" (AI) in official correspondence. (That’s not what SI means in my area of work, and reading this with that in mind is interesting.)

Clayton, a former SEC chairman, will chair the force. His vice chairs include FTC Chair Andrew Ferguson, the Pentagon's research chief, and the head of the Office of Personnel Management. The group reports directly to Trump and White House Chief of Staff Susie Wiles, and has 120 days to deliver a report on AI's risks and opportunities. Its charter reportedly says it will "develop plans for responding to SI-enabled threats to our society, while preventing overregulation and regulatory capture that would stifle innovation and competition."

The timing is the story. It came days after top AI executives signed a voluntary, non-binding "White House Accord on Super Intelligence," pledging to police themselves (LOL, what could possibly go wrong?!). When asked whether that accord was binding, Trump said, "I think it's morally binding." Yeah, yeah, we know how strong those morals are.

Follow the thread from the last month of this newsletter. Dario Amodei called for a coordinated industry slowdown. Then came the leaked Anthropic IPO prospectus warning of "existential risk." Now the federal government has planted its flag, and notably, Clayton publicly rejected Amodei's slowdown argument, telling CNBC "the risk of not being first is high." So within a few weeks we've gone from the industry saying "slow down" to the government saying "speed up, so we beat China," with the same small circle of people deciding what happens next.

Between the FTC's investigation into the top AI labs, the industry's voluntary self-policing accord, and now a task force reporting straight to the president, the rules governing the most consequential technology of our era will be written by a very small group, largely behind closed doors, explicitly structured to "prevent overregulation." Whatever comes next runs through those few hands.

Watch who is, and isn't, in the room. A task force built to coordinate the government's engagement with "consumers, public interest groups, critical infrastructure, and AI companies" will reveal its priorities by whose concerns actually shape the 120-day report. Watch also how "preventing regulatory capture" squares with a structure that concentrates decision-making among a handful of officials and the companies they're coordinating with. Those can be in tension.

The rules written by this small group will eventually govern the AI tools that show up in healthcare: in documentation systems, clinical decision support, and the devices our patients use. When policy for a technology this consequential gets concentrated among a few people explicitly tasked with not slowing it down, the professions on the receiving end, ours included, have even more reason to be AI-literate. Not because we'll be in that room. We won't. But because the more we understand how these systems work and how they're being governed, the better we can advocate for our patients when the downstream rules arrive at our clinics, ready or not.

And there's a quieter throughline connecting this to the top of today's issue. Both stories are about what happens when powerful institutions are trusted to police themselves, and about whether accountability actually reaches the people it's supposed to protect. In both cases, the public paying attention may be the only real check there is.

Back in issue #25, I broke down the research showing that AI-writing detectors don't reliably work, that they fail on the mixed human-and-AI writing that describes nearly every real submission, and that they falsely flag non-native English speakers and underrepresented students at alarming rates. My "what to watch for" was whether institutions would move away from detection toward process-based, trust-based assessment. A few months later, the shift has accelerated, and the courts have now entered the picture.

The institutional exodus grew dramatically. In March, I could point to about a dozen major universities that had disabled Turnitin's AI detection. As of this fall, that number is reportedly 25 or more major institutions, with some trackers counting 60-plus, including MIT, Yale, NYU, Johns Hopkins, and UC Berkeley, while UCLA declined to adopt it at all after reviewing the accuracy data. Crucially, most are keeping traditional plagiarism and similarity checking and dropping only the AI score, because they've concluded the two tools have genuinely different reliability.

And the biggest development, the one that should get every program's attention: this moved from policy to the courtroom. A New York court reversed a student's expulsion that was based on a Turnitin false positive. Students have begun filing lawsuits over false accusations. The reason the math makes this inevitable is worth restating plainly: Vanderbilt calculated that even Turnitin's claimed 1% false-positive rate, across its 75,000 annual submissions, means roughly 750 students wrongly accused every year. And independent studies keep landing well above that 1% claim, especially for the exact students I flagged in issue #25.

Everything I argued in issue #25 has strengthened, not softened. The tools are being abandoned, the courts are siding with falsely accused students, and the field is moving toward the exact alternative I pointed to: assessing the process (drafts, reasoning, revisions) rather than policing the finished product. For OT programs still relying on detection, this is now not just an ethics problem and a pedagogy problem. It's a legal exposure problem. Liability, if you will. The better path was always literacy over surveillance. Now there's case law starting to say so too.

(I'll keep following this and report on it periodically.)

Our rotating spotlight on where AI is showing up in the tools built for our field.

This week's headlines were heavy on who controls AI. Let's ground it back in where AI actually touches our patients: AI-guided telerehabilitation.

I presented on this earlier this year during a neurorehab symposium.

These are remote therapy platforms that use AI to personalize and monitor a patient's rehab program without a therapist physically present for every session. The AI adjusts task difficulty in real time based on how the patient performs, so the "just-right challenge" updates itself between visits instead of waiting for the next in-person appointment.

In a randomized controlled trial, Kim et al. (2025) found that self-guided, AI-driven cognitive telerehabilitation matched therapist-supervised outcomes in subacute stroke patients, with the AI adapting task difficulty in real time. Read that carefully: matched outcomes, not replacing the therapist. The patients still had a care team. The AI extended the reach of that team into the hours and days between sessions.

This is the neuroplasticity principle we've come back to all year, in action. Recovery is driven by sufficient, appropriately-challenging repetition. The clinic hour is only a sliver of a patient's week. A tool that extends well-calibrated practice into the home means more repetitions, more opportunities for experience-dependent plasticity, and potentially better outcomes, especially for patients who can't easily travel to frequent in-person sessions. For rural patients, homebound patients, and anyone with transportation or scheduling barriers, that reach matters enormously.

The promise is access and intensity. The risk is substitution creep, a drift from "AI extends the therapist's reach" to "AI replaces the therapist to cut costs." The Kim study matched outcomes with a care team in the loop. That's the design worth defending. An AI that adapts task difficulty is a genuinely useful tool; an AI deployed to justify cutting skilled therapy hours is different. The OT who understands the difference is the one who can advocate for the version that actually serves patients.

This issue is, from top to bottom, about leadership failures, institutions and leaders who chose convenience, self-protection, or speed over the people they were responsible for.

So let me turn that around and ask the harder, more useful question: what does it take to lead ethically when the environment around us rewards the opposite?

Because that's the situation most of us are actually in. Many of us lead something: a clinic, a classroom, a department, a team, a program—even if "leader" isn't in your title. And you're doing it in a moment when the examples at the top are not exactly inspiring.

Ethical leadership is easy when the incentives line up with doing the right thing. It only becomes leadership when they don't.

Here's what I think it actually requires, drawn from watching this week's stories play out:

The institutions that failed Jane Doe, and the governance structures being built around AI, share a pattern: accountability flows downward, onto the least powerful, while those at the top are shielded. Ethical leadership inverts that. It means holding yourself to the standard you set for others, and being answerable to the people you have power over, not just the people you answer to.

Every story this past month, from Amodei's pacing essay to the Super Intelligence Force's "risk of not being first," is about the pressure to move fast and skip the careful part. In our own work, the ethical move is often the slower one: the extra verification, the harder conversation, the "let me make sure this is right before we sign it." Leading ethically sometimes means absorbing the cost of being the person who won't just go along, or won’t be first at something.

The survivors going public, the researchers resigning and speaking out—these are costly acts of truth-telling against institutional pressure to stay quiet. You won't always face stakes that high. But you will face smaller versions constantly: naming a problem in a meeting, disclosing a limitation, admitting you changed your mind (I did that myself two weeks ago, revisiting praise I'd given a company earlier this year). Ethical leadership is built in those small, unglamorous moments of choosing honest over easy.

Our AOTA Code of Ethics names fidelity and justice as core: treating people fairly and keeping faith with those who depend on us. Those aren't abstract virtues. They're a description of what it looks like to lead well when leading well is inconvenient. The timeline we're in makes that harder. It also makes it matter more. The leaders worth following in an unethical moment are the ones who stay ethical anyway, especially when no one's forcing them to.

That's a heavy issue, I know. But if there's a thread running through all of it, it's this: in a moment when powerful institutions keep choosing to police themselves, the people paying attention, asking questions, and refusing to look away are doing something that matters. That's as true for sexual-violence accountability as it is for who gets to write the rules on AI, and as true for how we treat a falsely-flagged student.

Stay curious. Stay literate. And lead well where you are.

See you next week.

If this is your first article, thank you so much for being here! As I build this, I want to make sure it’s helpful to you. Please comment or email with any feedback or suggestions! And if you've found this useful, the most valuable thing you can do is forward it to one colleague who still thinks they can sit this out.

What would you like to read more of? Please let me know via email!

Until next time,

Pooja A. Patel, DrOT, OTR/L, BCG

learn. innovate. empower.