Voters will go to the polls soon in states and municipalities to decide ballot issues of significance for individual liberty, limited government, and sound public administration. Some highlights (via Ballotpedia and Bolts): 

Front and center for tax-watchers is the proposed California Proposition 40 wealth tax, criticized here in numerous pieces, including this policy analysis. Unusually steep by international standards at 5 percent, the measure would also defy widespread notions of legal regularity by reaching back retroactively to tax assets as of January 1, 2026, of persons who’d already moved out of state before the law’s enactment. Bad stuff.

Colorado’s Amendment 87 would replace the state’s flat 4.4 percent income tax with a graduated tax; it faces a counter-proposition that would freeze the existing flat rate. Iowa, Missouri, North Carolina, and Washington will all consider measures to freeze, roll back, or discourage increases in income tax rates. Adam Michel lately discussed the success of state flat taxes. 

Read the rest of this post →

Cato Surveillance Week 2026 featured a panel that included my colleague and Cato Research Fellow Nicholas Anthony. As is always the case with such expert panels, there are inevitably more questions than we have time to address in an hour. Moreover, because the reach of the Bank Secrecy Act (BSA) is so vast, I wanted to follow up with Nick on some of the issues we covered and some we didn’t have a chance to get into depth on. What follows is the transcript of my chat with Nick, edited for clarity. 

Q. Folks may not be as familiar with the Bank Secrecy Act as other laws. Can you give them a brief crash course?

The Bank Secrecy Act was enacted in 1970 to require banks and other financial institutions to monitor customers and report certain transactions to the government. It started as an attempt to go after tax evaders, but it has since been expanded to go after drug dealers, terrorists, and immigrants (and everything in between). 

Read the rest of this post →

More people continue to write of the need for Europe to become less dependent upon the United States for security, to provide for its own defense:

“The weaker the transatlantic relationship becomes, the more Europe needs other countries’ export markets, raw materials, defense industrial capacity, and solidarity against coercion—and the less able it is to convince those countries of its worth as a partner. Europe needs a better offer, grounded in its own capabilities rather than those of its mercurial ally.”

Read the rest of this post →

Economic reform doesn’t usually make for a good movie, except in the case of the “Chicago Boys,” a group of UChicago-trained Chilean economists who instituted neoliberal reforms in concert with the Pinochet dictatorship. 

New research examines the long-run effects of these reforms. By comparing 

Chile’s economic performance with an estimate of how its economy might have developed without the reforms … [the researchers found that the] 2019 real GDP per capita was approximately $4,012 higher than it would otherwise have been, a gain of about 41 percent relative to the estimated alternative. 

Read the rest of this post →

For the better part of the year, artificial intelligence (AI) policy has, reasonably, focused on establishing guardrails to mitigate some of the biggest risks in developing and deploying frontier AI models. This year has seen a rapid expansion in model capabilities, with the release of the first models capable of conducting advanced cyberattacks. We have also witnessed the first instances of models becoming advanced enough to outsmart the humans stress-testing them, raising some concerns about potential loss of control over frontier AI. In response, policymakers have enacted measures to mitigate some of the risks associated with the development and use of these advanced models. From the White House evaluation framework to the Joint Commitment on Frontier Responsibilities, rules setting up the AI industry’s duties and responsibilities have progressively increased.

However, while the industry faces increasingly intense scrutiny, there is worrisome vagueness about the government’s role and authority as a regulator. 

Read the rest of this post →

Documents obtained by Cato via the Freedom of Information Act (FOIA) reveal that more than a year before the 2026 midterms, the FBI was quietly building two election-related capabilities at once. The first was a set of internal “triggers” that told field agents when they could investigate AI-generated or “synthetic” political content — rules the Bureau is still keeping secret. The second was an apparently new headquarters post called the “FBI Election Executive,” who stood up in mid-2025 to “handle preparations for the 2026 midterm cycle.” In the months since that “Election Executive” position became public, nearly every public trace of the official who holds it has disappeared — her LinkedIn profile has been taken down, and there is no findable record of her on the FBI’s own website.

The FOIA Request and the Revealing Non-Answer

On February 27, 2026, I filed a FOIA request seeking six specific categories of records about FBI election-security activity conducted under the direction of the Office of the Director of National Intelligence (ODNI) — including the January 28, 2026, FBI raid on the Fulton County, Georgia, election office, at which Director of National Intelligence Tulsi Gabbard was personally present and reportedly put President Trump on the phone with the agents executing the search.

On October 13, 2026, the FBI finally responded to my FOIA request. It reviewed 41 pages, released 34, and withheld 7 in full. Not one of the 34 released pages answers any of my six questions. There is no ODNI tasking record, no 2011 ODNI-FBI memorandum of understanding, no legal analysis of ODNI’s authority to direct FBI field operations, no Fulton County coordination record, and no congressional-notification record. Nor did the FBI issue a “Glomar” response formally refusing to confirm or deny that such records exist. 

The categories I asked about are simply not addressed, which strongly suggests they were never searched. That failure is the subject of an administrative appeal I am filing this week.

Instead, the FBI released an internal Office of General Counsel (OGC) email chain on a different subject altogether — and that disclosure has implications for the 2026 midterms and the 2028 presidential contest.

Secret “Triggers” for Investigating Political Speech

The released emails document an FBI effort, launched in the fall of 2025, to write formal legal guidance for Special Agents in Charge (SACs) on when the Bureau may investigate AI-generated or “synthetic” media during the election cycle. The anchor document, dated September 30, 2025 — more than thirteen months before Election Day — is a background paper titled “AI in the 2026 Election Space.”

Interestingly, the unredacted portions are notably protective of the First Amendment, at least rhetorically. 

The paper states that synthetic media is “generally considered protected free speech,” that “the FBI may not take investigative action based solely on the exercise of First Amendment rights or on the race, ethnicity, gender, national origin, religion, disability, sexual orientation, or gender identity of the subject,” and that every investigation must rest on an “authorized purpose” under the Attorney General’s Guidelines and the Domestic Investigations and Operations Guide (DIOG). 

However, that “authorized purpose” is something Bureau and Justice Department officials essentially get to define themselves, a practice that has led to improper First Amendment-related investigations at scale, as documented by the Government Accountability Office (GAO) earlier this year.

But the sentence that matters most — the one listing the “general triggers” for when agents may review synthetic political content — is blacked out under FOIA Exemption 7(E), which shields law-enforcement techniques.

So, the public is told the FBI will respect the First Amendment but is not told the precise circumstances under which it has decided it need not do so. 

A promise with the operative terms redacted is not transparency; it is an assurance the public cannot verify, much less count on, as the GAO investigation I noted above revealed. Those triggers and the final SAC guidance built around them are the subject of a separate FOIA request that I will also file this week.

A New “Election Executive” — Ordered from the Top

The same email chain reveals that the guidance was commissioned at the very top of the Bureau. Then-Deputy Director Dan Bongino “requested an overview on AI in the election space,” the records show, and directed that OGC draft companion legal guidance “for all ADICs and SACs,” on a compressed timeline, “ideally out within mid-November.” 

Coordinating the work was a figure identified in the records as the “FBI Election Executive” — Kellie M. Hardiman, working out of the Office of the Deputy Director and, in her own words, “assigned to the DD’s office to handle preparations for the 2026 midterm cycle.”

Two features of that position deserve scrutiny. First, its novelty. The title “FBI Election Executive” appears nowhere in the public record before 2025. Hardiman’s own background is in intelligence, not elections: an independent, dated source — her speaker biography for the 2025 Law Enforcement Innovation Summit — identifies her as Deputy Assistant Director of the Intelligence Branch of the FBI’s Criminal Investigative Division, with a career spanning counterterrorism, counterintelligence, and a detail to the CIA. That 2025 bio makes no mention of elections. She appears to have taken the election role only in mid-2025 — contemporaneous with the secret AI-triggers workstream.

Asked by reporters in February 2026 when the position was created, the FBI would not say, asserting only that the role “is not a new role” — while naming no predecessor and providing no date. Pressed again by Politico, the government again declined to give a date, with the White House offering: “We don’t have anything additional to provide at this time.” A genuinely pre-existing position would have a creation date readily at hand.

The Vanishing Paper Trail

Here is where the story takes a very interesting turn. 

In February 2026, Hardiman signed correspondence as “FBI Election Executive,” inviting chief state election officials across the country to a briefing on the midterms. The outreach drew bipartisan unease; one invited official told NBC News the invitation was “unusual and unexpected” and that, of Hardiman, “no one has heard of this person.” At the time, reporters could at least find her: NBC cited a LinkedIn profile reflecting her appointment.

No longer. Between the February 6, 2026, NBC News story and October 19, 2026, that LinkedIn profile was removed from public view: the URL NBC cited now redirects to a dead end, and a related URL for the same account returns a “404” not-found page. The Internet Archive holds no capture of either, so there is no preserved public record of what it said. And a search of the FBI’s own website for “Kellie Hardiman” returns no result identifying her, even though the Bureau routinely publishes announcements of comparable senior appointments. The only public biography that survives is the one on a private conference vendor’s site — and it predates her election role and never mentions it.

The LinkedIn takedown is confirmed. The FBI website’s absence cannot, by itself, distinguish between a page that was removed and one that never existed. What is not in dispute is the net effect: the official the FBI publicly deployed to coordinate its election-year engagement with all fifty states has, within months, become nearly impossible to find in any government-controlled or self-published source. Whether by design or by neglect, that is the opposite of accountability.

The Statute in the File

One other document rounded out the release: a copy of the TAKE IT DOWN Act (Pub. L. 119–12), the nonconsensual-intimate-imagery and “digital forgery” law that Sen. Ted Cruz (R‑TX) and Sen. Amy Klobuchar (D‑MN) shepherded to enactment on May 19, 2025, and that the FTC began enforcing in May 2026. An OGC lawyer flagged it as “the one new legal development” worth noting to SACs — the reasons, naturally, redacted. It is a reminder that the Bureau’s interest in synthetic media is real and has a legitimate statutory basis. The problem is not that the FBI is thinking about deepfakes; it is that it is building secret rules about political speech and staffing a secret-ish office to run them.

Why It Matters

Strip away the FOIA mechanics and the sequence is this: in the year before a midterm election, the FBI’s political leadership ordered up secret rules governing when agents may investigate political speech, created an apparently new headquarters position to manage the Bureau’s election-year footprint, pointed that position at all fifty states’ election officials — and then the public record of the person running it thinned to almost nothing. Each piece might be explained away. Together, they describe an election-year expansion of federal investigative attention to political content, conducted with the operative details withheld and the responsible official rendered close to invisible.

The FBI may well have good answers. Congress should compel the Bureau to give them, on the record, with dates and with the “triggers” unredacted, so we can all judge whether the Bureau’s conduct in this area passes constitutional muster.