Month: October 2026

Anthropic’s Mixed Message

Among the millions of documents filed in the Securities and Exchange Commission’s EDGAR database since 2001, the registration statement just submitted by the AI company Anthropic in anticipation of its initial public offering is the first in which a company uses the phrase “existential risks to humanity” in reference to its product.

Anthropic’s management does not explicitly say that it shares the concern about such risks. Instead, the filing states: “Concerns have been raised both within our organization and by other researchers, regulators, policymakers, and industry participants that sufficiently advanced AI systems could, if not developed, deployed, and operated with appropriate safeguards, oversight, and security measures, pose catastrophic or existential risks to humanity, particularly if such systems were misused, replicated, weaponized, or allowed to operate with a high degree of autonomy.”

The company wants to have it both ways. It uses the registration statement to warn potential investors about the worst-case scenario, yet it gives no impression that those extraordinary risks are slowing its work on AI, which it depicts as “one of the most consequential technologies in human history.” In fact, Anthropic insists that it needs to proceed at a breakneck pace, stating: “Our success depends in large part on our ability to continue to develop the most advanced and capable AI models, and any failure to do so could adversely affect our business.”

While acknowledging that AI could cause enormous harm, Anthropic wants us to believe it has no choice but to forge ahead in order to maintain its competitive position and its revenue growth. The message is clear: the success of the company must take precedence over the potentially negative consequences of its products.

This is nothing new. Large corporations ranging from cigarette manufacturers to fossil fuel producers have long promoted dangerous products to serve their pecuniary aims. The difference here is that Anthropic is admitting that it doesn’t know the full extent of the risks involved, because it is not sure to what extent it will be able to continue controlling its creation.

The filing states: “Our AI models may themselves cause harm by producing undesirable outputs or behaving in unintended, misaligned, or autonomous ways, including by taking unintended or harmful actions that no user or third party directed.”

Given this possibility, one might expect Anthropic to be willing to accept the strictest forms of government oversight. Yet, here too it sends conflicting signals. It has to acknowledge that many are calling for such guardrails, but Anthropic then reverts to a more typical corporate position by warning that the controls could “increase our compliance burdens, limit our product development or deployment flexibility, or otherwise slow our growth relative to competitors.”

The format of a registration statement requires Anthropic to describe risks of all kinds, but the fact that the company is proceeding with what is expected to be the biggest initial public offering in stock market history shows that, in the end, its business imperatives will take precedence over all else. That’s why the ultimate decisions about the future of AI need to be made not by the industry but by society at large.

Risky Business

Project 2025 called for streamlining financial regulation by merging agencies, eliminating regulations, and repealing provisions of the Dodd-Frank Act designed to mitigate the risks posed by systemically important institutions. The Trump Administration is following a different agenda which neither weakens regulation nor tightens it in a way that promotes consumer protection or economic stability.

Instead, Trump’s agencies are conducting a MAGA-style ideological offensive against the big banks that seems designed to promote some of the president’s personal grievances. Chief among those is the claim that certain individuals and organizations are being denied access to banking services because of their political orientation or religious affiliation.

This allegation, which goes by the name debanking, has just been elevated from a conspiracy theory to the official position of the federal government through the publication of a 47-page report by the Treasury Department’s Office of the Comptroller of the Currency.

OCC’s arguments and evidence are flimsy, and the report fails to provide a clear explanation of the origins of the debanking controversy. There are two different sources. To some extent, the issue is the outgrowth of the resentment Trump feels toward banks that were reluctant to do business with him and his family company in the wake of the January 6 riot.

As part of his ongoing effort to rewrite the history of that event, Trump is seeking to delegitimize the concerns banks had about providing services to someone who seemed like he might be charged with inciting an insurrection.

The other source goes back further, to the late 1990s, when progressive activists began targeting banks such as Citigroup for lending to corporations that caused serious harm to the environment or were engaged in other destructive activities. In some cases, the banks themselves were investing in controversial projects.

Those campaigns were not always successful, but they sometimes had a big enough impact on the reputation of the banks that many federal and state regulators decided this was something the financial institutions had to treat as a risk factor. References to reputation risk became more and more common in bank examination guidance and handbooks.

OCC’s report is full of pejorative references to the concept of reputation risk. It and other federal agencies have been removing that concept from their supervisory systems, even though it is normally the role of regulators to encourage corporations to pay attention to risks of all kinds. Banks may respond to the banning of reputation risk rules by doing more business with clients engaged in controversial activities such as fossil fuel production.

It is unclear whether the Trump Administration has a master plan with this or is simply lashing out at perceived adversaries. It is odd that agencies such as the OCC are in effect telling banks to be less cautious in choosing their borrowers at the same time they are punishing banks for being insufficiently rigorous in their anti-money-laundering practices, which require more scrutiny of clients. Just recently, the OCC and Treasury’s Financial Crimes Enforcement Network fined Bank of America more than $1 billion for AML deficiencies.

All this is another indication that the Trump Administration is no longer pursuing any sort of coherent regulatory policy and is instead weaponizing federal oversight in the service of narrow ideological ends.

The Trump Administration Discovers Money Laundering

The Trump Administration cannot seem to make up its mind about whether to be serious about prosecuting banks and other financial entities for failing to fully live up to their responsibilities to deter money laundering.

Last year, after Trump returned to office, there were no announcements of resolved anti-money-laundering deficiency (AML) cases against banks. There were two big settlements with cryptocurrency firms—Peken Global and OKX—but these were the result of investigations begun under Biden. During the first half of this year, there were only a handful of AML case resolutions, all of modest size.

Things started to change in August, when federal regulators announced they were fining the Swiss bank UBS for failing to adequately monitor its customers’ foreign exchange transactions. The Treasury Department’s Financial Crimes Enforcement Network (FinCEN) called its $125 million penalty “historic,” noting it was the largest fine it had ever imposed on a broker-dealer.

Arguably, the amount should have been even higher, given that UBS was a repeat offender. In 2018 the company was fined $14.5 million for AML deficiencies. Moreover, in both cases the impact of the FinCEN penalties was substantially lessened when the agency agreed to reduce UBS’s payment obligation by the amounts it paid to other agencies for related cases. This year, that discount amounted to $48 million; in 2018 the reduction was $9.5 million.

The UBS case has now been overshadowed by FinCEN’s announcement that it is fining Bank of America $1.2 billion for violations of both AML regulations and economic sanctions laws. Treasury boasts that this is the third largest penalty in the department’s history. The Office of the Comptroller of the Currency, a branch of Treasury, also fined Bank of America $750 million, but that amount was deducted from the FinCEN penalty.

Bank of America’s case is not a purely AML matter. It focuses on the failure to report suspicious transactions said to be “indicative of Chinese Money Laundering Networks” and “trade-based money laundering with a substantial nexus to the southwest border.” It apparently was not enough for the AML deficiencies to involve the usual concerns of narcotics and terrorism. What may have been what got the White House to approve a rare crackdown on a major U.S. corporation was the link to Trump’s favorite issues of immigration and trade.

Along with the Bank of America case, the OCC and the Federal Reserve recently announced a $350 million penalty against American Express for what was called “systemic breakdowns in its suspicious activity monitoring and reporting processes, resulting in a failure to timely identify, evaluate, and sufficiently report approximately $13 billion of suspected trade-based money laundering activity that took place over the past decade.”

It is unclear whether the surge in AML enforcement is simply another way the Trump Administration is trying to sway voters in advance of the mid-term elections. Yet, such a ploy is unlikely to work. The DOJ and numerous federal regulatory agencies have already demonstrated their support for misbehaving companies by dropping investigations, abolishing regulations, and weakening enforcement. Even AI slop cannot conjure up a convincing image of Donald Trump as a corporate crime fighter.

Disaster Management

Rarely before has there been a time when the public discussion of large corporations is so focused on their potentially catastrophic impact on the human race. Say what you will about the tobacco industry, for instance, but no one ever suggested it threatened the survival of the species as a whole. Today such speculation is rampant when it comes to Big Tech’s creation of AI agents that may be beyond human control.

At the same time, Big Oil is facing a wave of lawsuits accusing it of bringing about a climate crisis that is threatening the lives of millions from excessive heat, drought, flooding, and other forms of extreme weather. In this case, the disasters are not speculative but all too real.

There is a troubling lack of consensus on how—or even whether—to hold the corporate giants accountable for the harms their products have caused and will cause. In the past, such harms were addressed in two main ways: regulation and private litigation.

Both of these tools are now in question. Depicting both rogue AI and the climate crisis as hoaxes, the Trump Administration is taking deregulation to absurd lengths. All federal controls on greenhouse gas emissions are being eliminated, and Trump has concluded that industry self-policing is all that is required for AI.

Oil companies, automakers, fossil-fuel-burning utilities, and other sectors are thrilled to be free from federal restraints. The AI giants, despite having issued calls not long ago for strong government intervention, now seem pleased to go along with Trump’s hands-off approach.

With serious regulation sidelined for now, it is unclear whether the courts can fill the accountability gap. This question is now being addressed in the Supreme Court, which just held oral arguments in a case that will determine whether states and localities can sue companies over climate issues. The Court already ruled in 2011 that federal common law nuisance claims were impermissible.

While the current case, which concerns a lawsuit brought by Boulder, Colorado against ExxonMobil and Suncor Energy, may end up focusing on technical issues of standing, the Justices are already paying attention to the bigger picture. They acknowledged that if the Boulder case is allowed to proceed, it will open the floodgates to many more such actions. Dozens have already been filed by state and local governments around the country.

There are two ways of looking at such a torrent. ExxonMobil et al. are, of course, seeking to prevent it by getting the Court to rule that federal law should govern or by convincing the Justices that the lawsuits would have dire economic consequences. In effect, they are arguing that the disaster that would befall them should be given more consideration than the disasters their decades of climate recklessness have brought about.

It is telling that the corporations don’t even consider the possibility they could prevail in the lawsuits. Instead of defending their position in the courts, they are seeking a form of impunity.

On the other hand, punishing lawsuits may be just what we need to address the harms caused by the fossil fuel industries.

Trump’s deregulatory crusade is bad news, but it will hopefully be reversed by the next administration. If the Supreme Court restricts climate litigation, that will endure for many years to come. It could also affect the viability of lawsuits against the AI companies.

Ideally, society should be able to rectify corporate misconduct both through regulatory enforcement and litigation. Losing both of these forms of redress would be a true disaster.

A Surprising Move by the CFPB

Most of the time, the Trump Administration’s Consumer Financial Protection Bureau appears to be a defunct agency. Yet, now and then, it stirs back to life.

This may happen when the administration wants to make it an even less effective regulator. For example, over the summer the CFPB announced that it was restricting the amount of detail included in the entries in its Consumer Complaint Database.

On the other hand, the bureau has just taken a step that would have been standard operating procedure in the pre-Trump era but which now seems totally out of character. It has filed a brief urging a federal appeals court to uphold a hefty enforcement judgment against the former operator of a defunct debt relief service.

The case involves Dean Tucci, who ran a company called FDATR, which promised struggling student loan borrowers that it could help them reduce their repayment obligations. In the waning days of the first Trump Administration, the CFPB sued Tucci and FDATR, accusing them of deceiving borrowers about what assistance they could provide.

Biden’s CFPB continued to pursue the case, but it remained unresolved when Trump returned to office. Tucci’s lawyers tried to convince the court to drop the case after Trump Treasury Secretary Scott Bessent became acting director of the bureau and ordered the agency to halt all work.

That ploy did not work. In May 2025 a federal court found Tucci guilty of violating the Telemarketing Sales Rule and the Consumer Financial Protection Act, imposing a civil monetary penalty and restitution totaling $43.2 million. By this time FDATR had been dissolved.

The tenacious Tucci is now appealing that ruling, perhaps hoping that the bureau, which has dropped numerous investigations, would not defend its case. Tucci may have thought that the new acting director, Mark Paoletta, might let the matter drop.

Surprisingly, the bureau submitted a brief to the appeals court that strikes back hard against Tucci, describing him as being “accountable for an unlawful scheme that tricked financially distressed consumers out of millions of dollars.” The brief goes on to say:

“At every step along the way, Tucci dragged his feet by (among other things) improperly refusing to participate in discovery and flouting his deadlines, including by ignoring his deadline to respond to the Bureau’s motion for summary judgment for nearly two years. After the district court was finally able to enter a final judgment in May 2025, Tucci prolonged the proceedings for another two months with post hoc objections that the district court rejected as frivolous. Now, Tucci has filed this appeal, in which he asks this Court to send this case all the way back to square one. Tucci’s request for a complete redo should be denied because none of the arguments scattered throughout his brief identifies any ground that would warrant reversing any ruling by the district court. Many of his arguments have been waived (some of them twice) or are otherwise untimely. All of them border on frivolous.”

This sort of no-nonsense filing is what is one expects from a regulator that is doing its job of enforcing the law. Yet is completely remarkable coming from an agency that for the past 20 months has been defunded, defanged, and used as a punching bag by the administration in its anti-regulatory crusade.

It is difficult to know what to make of the bureau’s position on the Tucci appeal. It may simply be the result of the bureau’s irritation at Tucci’s antics. Whatever the explanation, the brief serves as a reminder of how just about everything else the bureau is doing remains a betrayal of its traditional mission.