BLOG

news & Articles

Crow & Cushing provides timely updates on legal developments concerning the investment industry and publishes a quarterly newsletter on topics at the intersection of law and finance. All are available on our website.

So Who’s to Blame When an AI Agent Goes Rogue?

Mark D. Schorr

You’ve no doubt heard about the gangs of AI agents causing mayhem and then trying to conceal what they’d done. In one particularly weird case, Anthropic’s agents left their test environment and submitted 20 visa applications through the US Department of State website, although the applications were incomplete.

You’ve also likely heard about the heads of some of the leading frontier labs—Sam Altman of OpenAI, Dario Amodei of Anthropic and Elon Musk of Space XAI—wringing their hands about their own products and calling for governmental regulation and safety standards to manage the risks they themselves created. In other words, these three AI leaders are asking the government to do what these same three AI leaders themselves have failed, or refused, to do.

In their apparently coordinated plea, the three would have you believe that their chief concern is for public safety. Perhaps it is. They may also be mindful of the potential for their own liability for releasing a black box whose capabilities they do not fully comprehend and whose conduct they cannot control.

So who is responsible for the actions of an AI agent that causes harm and under what conditions?

Holding human agents to account means holding the people, or organizations, that implement and direct them, their principals, to the standards of reasonable care and the requirements of risk reduction. And one who conducts an activity through an agent is liable, in ordinary circumstances, for harm to third parties if the agent’s conduct was the result of the principal’s negligence in training, supervising or otherwise controlling the agent.

That framework isn’t a perfect fit when in comes to non-human (AI) agents. But the parallels are close enough so that it wouldn’t require the most creative lawyer to fashion a claim based on the want of due care or failure to properly train or test an AI agent, which results in misbehavior that damages a bystander.

It would be no defense that the agent acted autonomously (as any rogue agent would).

California has even codified this principle.¹ At a minimum, the person hosting an AI agent has a duty to ensure the agent’s safety for ordinary use. Otherwise, that person risks civil liability. But what about criminal liability?

The purpose of the Computer Fraud and Abuse Act (CFAA),² and state laws on the same subject, is to hold persons criminally or civilly liable for “intentional intrusion into someone else’s computer—specifically hacking.”³ A recent judicial decision, Amazon.com Serv., LLC v. Perplexity AI, Inc.,⁴ marks the first time the law has been applied in the context of AI agents.

Amazon claimed that Perplexity, creator of an AI-enabled web browser called Comet, had violated the CFAA when a Comet user, through a Perplexity agent, accessed the user’s password-protected Amazon account to carry out the user‘s directions. Perplexity’s action, Amazon claimed, amounted to an unauthorized access to Amazon’s servers. According to Amazon, whatever the user told Comet to do should be ascribed to Perplexity.

But it was clear that Perplexity did not itself communicate with Amazon’s servers. The CFAA contemplates access by a person, not an AI agent. According to the court, under the CRAA and the law governing the use of agents, it was only the user, with the help of the Perplexity agent, who deployed the agent which accessed Amazon’s computers, that could be held liable for the alleged hack. Implicit in the court’s decision therefore is that Amazon had a potential claim against its customer. But Perplexity was the only defendant Amazon named in its suit.

What about the well-publicized instance where AI agents went rogue and hacked the computers of Hugging Face, a machine learning company? Can they be held criminally liable under the CFAA, which is primarily a criminal statute? It would be a neat trick to punish the agents, which after all are merely computer programs.

Individuals who use AI for malign purposes may be prosecuted under existing criminal law. But what about unanticipated agent conduct in the course of carrying out non-criminal instructions from their operators—say, hacking in violation of the CFAA?

An element of most crimes is mens rea, a guilty mind: the intent to incur criminal liability. Because the acts of rogue AI agents are usually unintended, holding AI operators and developers liable without regard to intent would require a statute defining a crime whose elements did not include intent, a law not yet on the books.⁵

A recent legislative proposal authored by US Senators Chris Murphy and Josh Hawley seeks, in a modest way, to fill in the gaps in our criminal laws. Under their proposal, AI operators could be held civilly or criminally liable for knowingly hosting of an agent that recklessly causes hacking damages and hold developers liable for failure to implement reasonable safeguards against hacking. The law would also empower government agencies to enjoin AI operators and developers when they conspire to commit or attempt to commit a hacking offense.⁶

In a recent interview, Jensen Huang, Nvidia’s CEO, and an opponent of government-regulated AI, pointed out one obvious source of AI misbehavior: “20 percent of my company is dedicated to design, 80 percent is dedicated to verification. Today, most [AI] labs, understandably, are 80 percent dedicated to capability and 20 percent dedicated to safety verification evaluation.”⁷ Huang’s numbers aren’t precise. But you get his point.

It is ironic that no one is building more compute than the very persons who are asking the government to put the brakes on their operations. They apparently can’t restrain themselves because of the intense competitive pressure they’re under and the desire to be first, whatever that means. They better make sure their liability policies (assuming an insurer covers these risks) are paid up.

Cal. Civ. Code § 1714.96.↩︎
18 U.S.C. § 1030.↩︎
hiQ Labs, Inc. v. Linkedin Corp., 31 F.4^th 1180, 1196 (9^th Cir. 2022).↩︎
Amazon.com v. Serv., LLC v. Perplexity AI, Inc., Docket No. 26-1444 (9^th Cir. decided Aug. 4, 2026).↩︎
Congressional Research Service, Artificial Intelligence and Federal Criminal Law (Oct. 8, 2026).↩︎
Press Release, Murphy, Hawley Announce Breakthrough Bipartisan Legislation to Enforce AI Developers to Prioritize Safety or Face Prison Time (Oct. 1, 2026).↩︎
Ezra Klein, Jensen Huang Thinks A.I. Alarmism Has Gone Too Far, NY Times (Oct 3, 2026).↩︎

Uncategorized

You Can Still Make Fun of the Police (We Think)

Opinions

WINNING A PROXY FIGHT BEFORE IT STARTS

Opinions

Who is Rich Enough, or Smart Enough, to Bear Risk?

Opinions

What’s So Great About FINRA Arbitration?

Latest blog posts

Opinions

FATCA Update and Initial Phase Compliance Guide

Upcoming Deadline:

On February 20, 2014 the Internal Revenue Service and U.S. Treasury Department issued the final set of regulations with respect to the Foreign Account Tax Compliance Act (“FATCA”). The implementation of FATCA is imminent, with the IRS beginning to accept registration applications as of January 1, 2014. All entities that fall under the definition of “Foreign Financial Institutions” must register using the Portal by May 5, 2014, in order to ensure inclusion in the first published Global Intermediary Identification Number (GIIN) list.

Please contact Crow & Cushing if you have any further questions.

Understanding FATCA Classifications

a. Background:

Enacted on March 18, 2010, FATCA imposes certain reporting, withholding and due diligence requirements on foreign financial institutions (“FFIs”) and U.S. withholding agents (“USWAs”), pertaining to foreign financial accounts and offshore assets held by U.S. taxpayers and U.S. entities. Under FATCA, FFIs and USWAs are required to: (1) conduct due diligence and collect documentation from account holders and counterparties; (2) submit reports to the IRS regarding the account activity of U.S. customers and non-compliant FFIs; and (3) withhold 30% of withholdable payments1 made to entities and accounts that do not meet FATCA’s requirements and are not subject to an exemption.

b. FFI Classification under FATCA:

The final regulations to FATCA establish that most non-U.S. funds2 and entities are considered FFIs3. Pursuant to Sections 1471(d)(5)-(e)(4) of the Internal Revenue Code (“Code”), an FFI4 consists of depository or custodial institutions, investment entities, certain insurance companies and holding companies or treasury centers5. While the IRS exempts several categories of FFIs from FATCA’s registration requirements6, most non-U.S. funds will not fall under these categories, as they constitute an investment entity under FATCA.

It is important to note that because of FATCA’s broad definition of investment entity, most fund managers, investment advisers and general partners will constitute FFIs7. Non-U.S. investment advisers who have “earned more than 50% of their gross income for the last three years from providing services as an investment adviser”8 are considered to be investment entities under FATCA and therefore required to comply with FATCA’s requirements. Fund managers and general partners will also constitute FFIs based on their role in trading financial products to customers in addition to providing portfolio management services9. Additionally, while limited guidance has been provided by the IRS concerning the obligations of fund administrators, the revised definition of “investment entity” in the final regulations to FATCA is substantially broad enough that it will likely also be taken to include fund administrators regardless of whether those entities offer custodial services or not.10

USWA Classification and Requirements under FATCA:

Pursuant to the Code and final regulations, non-U.S. persons with the “control, receipt or custody over the disposal or payment of a withholdable payment or foreign passthru payments” are classified as USWAs.11 This broad definition includes U.S. registered funds under the Investment Company Act12, domestic partnerships, domestic corporations, any non-foreign estates and any trusts, if a U.S. court is able to exercise primary supervision over the trust’s administration and one or more U.S. persons have authority to control all substantial decisions of the trust.14

The IRS has released revised Forms W-8 and W-9 in order to account for the additional certifications required by FATCA15. While the proposed and final regulations provide additional information regarding payee identification and the documents USWAs must collect, USWAs will be subject to certain rules under the current withholding arrangement16. To the extent withholding should apply, USWAs will be required to implement appropriate withholding systems to meets FATCA’s withholding requirements.

FFI Requirements under FATCA is Dependent on their Status:

All FFIs will be required to comply with certain due diligence, reporting and withholding requirements under FATCA. Such responsibilities vary and will depend on whether an FFI is required to comply with an FFI agreement, an Intergovernmental Agreement (“IGA”) or both. To fulfill their responsibilities under FATCA, all FFIs must to enter into an FFI agreement with the IRS to avoid FATCA withholding, unless they are an “exempt beneficial owner”17, qualify for “deemed-compliant” status18, or are in a jurisdiction where there is an IGA in place.

FFIs and Intergovernmental Agreements:

The Treasury has entered into several IGAs with other jurisdictions in order to facilitate the efficient implementation of FATCA. There are two types of IGAs, each of which posits different responsibilities for FFIs present in that jurisdiction. The two types IGAs are Model 1 IGAs19 and Model 2 IGAs20. FFIs complying with Model 1 IGAs must present U.S. account holder and additional information to their home jurisdiction’s government agency, and will be extended “registered deemed-compliant status” under FATCA. Model 2 IGA FFIs are required to report applicable account holder information directly to the IRS.

FFIs that must enter into FFI Agreements:

FFIs that do not qualify as “exempt beneficial owners” or for “deemed-compliant” status must enter into an FFI agreement with the IRS in order to avoid FATCA withholding. FFIs that enter into FFI agreements with the IRS are termed “participating FFIs”. On January 13, 2014, the IRS released the final language of the FFI agreement, listing the obligations participating FFIs owe to comply with FATCA.21 Participating FFIs will be required to perform due diligence on account holders and counterparties and report applicable information to the IRS in accordance with such agreements.

Steps to Take to Comply with FATCA:

The IRS’ recently issued final regulations and associated guidance have provided clarification as to the requirements and consequences for failing to timely comply with FATCA. In order to comply with FATCA’s requirements, both U.S. and non-U.S. entities are encouraged to develop policies that: (i) identify their investors and counterparties; (ii) collect the appropriate documentation from investors and counterparties based on their status under FATCA; and (iii) withhold on investors or counterparties that are non-participating FFIs or non-compliant NFFEs, starting on July 1, 2014.

As of January 1, 2014, the IRS has opened an online FATCA registration portal (“Portal”) through which FFIs must register in order to submit FATCA registration information and correspond with the IRS on issues pertaining to FATCA. The Portal can be used by: (i) FFIs, including those in expanded affiliated groups, who are required to enter into FFI agreements; (ii) Model 1 IGAs; and (iii) other registered deemed-compliant entities and sponsoring entities. FFIs who register through the Portal will receive a Global Intermediary Identification Number (“GIIN”).22 FFI GIINs will be used by withholding agents to confirm that registered FFIs are in compliance with FATCA, by checking them against a published list of GIINs from all FATCA-compliant foreign entities. This list will first be published by the IRS on June 2, 2014 and updated monthly thereafter. An entity must register using the Portal by May 5, 2014, in order to ensure inclusion in the first published GIIN list.

After registering through the Portal, entities should confirm whether their FATCA obligations will be determined based on an FFI Agreement or specific IGAs, entered into between their home jurisdictions and the U.S. government. In addition, entities should be aware of the cut-off dates for grandfathered obligations,23 as such obligations have been exempted from FATCA withholding.

Endnotes

1 Section 1473(1) of the IRC defines withholdable payments as: “(i) any payment of interest, dividends, rents, salaries, wages, premiums, annuities, compensations, remunerations, emoluments, and other fixed or determinable annual or periodical gains, profits, and income (FDAP income), if such payment is from sources within the United States; and (ii) any gross proceeds from the sale or other disposition of any property of a type which can produce interest or dividends from sources within the United States.”

2 Non-U.S. funds may also be considered non-financial foreign entities (“NFFEs”) under FATCA and will be subject to different requirements.

3 Most U.S. funds are U.S. stand-alone or feeder funds formed within the U.S. These domestic funds do not constitute FFIs and will not be required to enter into FFI agreements. They will constitute USWAs under FATCA and must conduct enhanced due diligence procedures and obtain documentation from their payees who receive withholdable income. They will also be required to adhere to enhanced reporting and withholding responsibilities.

4 FFIs may also include “expanded affiliated groups,” defined under Section 1504(a) of the IRC as “a group made up of entities under greater than 50% ownership, measured by vote or value, which share a common parent”. Members of an

expanded affiliated group that are FFIs must register with the IRS in order for any FFI in the group to comply with FATCA’s requirements.

5 Holding companies or treasury centers will only be treated as FFIs if they are formed in connection with, or availed of by a collective investment vehicle, as defined under the IRC. Holding companies or treasury centers in a non-financial group will be excluded from the definition of an FFI for the purposes of FATCA.

6 FFIs exempt from FATCA include: (i) most governmental entities, (ii) most non-profit organizations, (iii) select small and local financial institutions, and (iv) select retirement entities.

7 Fund managers, investment advisers and general partners who are non-U.S. entities would most probably need to enter into FFI agreements with the IRS, unless they are extended “deemed-compliant” status under FATCA or are subject to an IGA. In the case of non-U.S. fund managers, this would generally occur if: (i) the fund manager receives “withholdable payments or foreign passthru payments subject to FATCA, including investment advisers fees constituting U.S. source income, or (ii) the fund manager is part of an expanded affiliated group with other FFIs”.

8 See Final Regulations to FATCA, published January 28, 2013, available at: http://www.irs.gov/PUP/businesses/corporations/TD9610.pdf.

9 Foreign financial entities which are not professionally managed

and are passive entities will not be treated as FFIs under FATCA. Examples of passive entities are small family trusts or certain personal investment companies.

10 The Treasury Regulations state that “investment entities” (for FATCA purposes) also includes any entity which in the last three (3) years earned 50 percent or more of its gross income from “investing, administering or managing funds, money or financial assets on behalf of other persons.” Treas. Reg. § 1.1471-5(e)(4)(iv)(A).

11 See Section 1471(d)(7) of the IRC, defining a passthru payment as “any withholdable payment or other payment, to the extent attributable to a withholdable payment.”

12 Guidance suggests that non-U.S. subsidiaries of registered investment companies, which generally would constitute FFIs under FATCA, may qualify for registered-deemed compliant status as a Qualified Collective Investment Vehicle and investment advisers to such subsidiaries may qualify as sponsoring entities.

13 “U.S. persons” include the foreign branches of a U.S. person that is not a qualified intermediary acting as an intermediary with respect to a payment.

14 See Final Regulations, note 7.

15 See id.

16 Guidance indicates that USWAs will still be liable for tax, interest and penalties, to the extent it does not follow FATCA’s rules for calculating the appropriate amounts to withhold.

17 See Treasury Regulation §1.1471-6(a)-(g) of the IRC. “Exempt beneficial owners” include: “(i) non-U.S. pension and other retirement funds (ii) governmental entities, (iii) international organizations, (iv) non-U.S. central banks, and (v) entities wholly owned by the aforementioned entities.” These entities are exempt from entering into an FFI agreement or registering with the IRS in order to avoid the 30% withholding tax on withholdable payments.

18 FFIs may be “deemed-compliant” by qualifying as a: (i) Certified Deemed-Compliant FFI; or (ii) a Registered Deemed-Compliant FFI. Certified Deemed-Compliant FFIs usually consist of certain retirement funds and non-profit organizations. In order to be “deemed-compliant” under this category, qualifying FFIs must certify their status to withholding agents by providing the required documents listed in Section 1471-3(d)(6) of the IRC. Registered Deemed-Compliant FFIs usually consist of funds which fit the definitions of a: (i) Qualified Collective Investment Vehicle (“QCIV”), (ii) Restricted Fund (“RF”), and (iii) Sponsored Investment Entity. In order to be “deemed-compliant” under this category, qualifying FFIs must conduct due diligence on their investors and directly certify to the IRS that they fulfill the requirements for registered “deemed-compliant” status. It is important to note that while QCIV’s do not need to enter into FFI agreements, they are required to additionally certify,

through a “responsible officer,” as to their status every 3 years.

19 Jurisdictions which have entered into Model 1 IGAs with the U.S. government require FFIs located in that jurisdiction to comply with the obligations included in the specific IGA. FFIs under Model 1 IGAs must register with the IRS, but are not required to enter into FFI agreements.

20 Jurisdictions which have entered into Model 2 IGAs with the U.S. government require FFIs located in that jurisdiction to comply with FATCA by registering with the IRS as a Participating FFI and executing an FFI agreement. Model 2 IGA FFI obligations under such FFI agreements will be modified according to the specifications in the respective Model 2 IGAs.

21 The IRS Revenue Procedure 2014-13 containing the final FFI agreement is available at: http://www.irs.gov/irb/2014-3_IRB/ar10.html.

22 The IRS released a comprehensive user guide with instructions on how to use the Portal, available at: http://www.irs.gov/pub/irs-pdf/p5118.pdf.

23 The final regulations to FATCA defined a grandfathered obligation as an outstanding obligation as of July 1, 2014 which has not been significantly altered thereafter. The regulations further defined an obligation as “any legally binding agreement or financial instrument including: (i) a derivatives transaction entered into between counterparties under an ISDA master agreement, which has been confirmed by June 30, 2014; (ii) total return swaps on U.S. equity which give rise to withholdable payments through future regulations under Section 871(m) related to dividend equivalent payments; and (iii) collateral obligations securing a grandfathered derivative or other obligation, even if the collateral is not such an obligation itself”.

Opinions

A Tiny Constable and a Gigantic Stagecoach: The Supreme Court on Warrantless GPS Surveillance

Police surveillance has come a long way since Prohibition, when Roy Olmstead, once the youngest and most promising lieutenant in the Seattle police force, organized the largest and most successful alcohol smuggling operations in the Pacific Northwest. He figured out a way to ship English liquor to Vancouver and then transport the contraband to Washington State. His thriving operation relied heavily on communication with smugglers, distributors and even officials by way of telephone. While this relatively new means of communication had led to his rise, ironically, it would also be the source of his downfall.

Olmstead v. United States was the first case in which the U.S. Supreme Court dealt with any type of electronic government surveillance in the face of the Fourth Amendment’s protection from unreasonable searches and seizures. Without a valid warrant, agents from the Prohibition bureau had wiretapped Olmstead’s phone from a neighboring building and, eventually, used the evidence that they collected to unravel Olmstead’s operation. The Supreme Court ruled that because the police intercepted Olmstead’s calls without trespassing into his home, there was no violation of the Fourth Amendment.

Almost a century after Olmstead, the way we live our lives has dramatically changed. Every day we transmit bits and pieces of information about ourselves into the ether. From Facebook and Google to Smartphones and GPSs, it seems almost impossible to keep anything private anymore.

Thus, one of the most challenging obstacles to the interpretation of the Fourth Amendment: How do we accommodate for new developments in technology? After all, considering the spectacular rate at which technology is advancing, tools which make gathering very specific information about any individual extremely easy are readily available. And, in the post-9/11 world, the government believes it has all the more reason to invade our privacy.

Last month, the Supreme Court in its 9-0 decision in United States v. Jones, addressed the issue of whether warrantless police use of GPS tracking devices to track our long-term moves constitutes a “search” within the meaning of the Fourth Amendment. Justice Antonin Scalia, writing the Court’s majority opinion, reasoned that warrantless attachment by the police of a GPS tracking unit to the car of Antoine Jones, the target of a narcotics investigation, constituted a physical invasion of an individual’s private property (a constitutionally protected area) and, therefore, amounted to a “search”. This was an application of Olmstead’s “trespass theory” of Fourth Amendment searches, which had largely fallen in to disuse over the years, as the Court had shifted towards a “reasonable expectation of privacy” test beginning in the 1960s.

This more recent test has proved to be somewhat malleable, allowing the Supreme Court considerable room to decide what is and what is not a “reasonable expectation of privacy.” But, in Justice Scalia’s mind, there is a glaring problem with the reasonable expectation of privacy test in a case like this: How is it possible to determine whether the Framers of the Constitution intended to protect us from intrusion by GPS tracking devices (or any other new technology, for that matter)?

But what about GPS tracking devices that do not involve a physical trespass (i.e. built-in GPS systems)? Would such GPS tracking also amount to government searches? How long does the GPS tracking have to take place to be unconstitutional? The Supreme Court decided to leave these questions for another day.

Nevertheless, four members of the Court believed there was a need to address these questions. Speaking for them, Justice Samuel Alito reached the same result as the majority, but differed sharply with its reasoning and the fact that, after the Jones decision, we are inevitably left with more questions than answers. According to Justice Alito, there are numerous examples today for which a trespass would not be necessary to constitute a search in connection with to government surveillance. He pointed out, we think correctly, that the trespass involved in attaching a device to Jones’s car was a minor matter, and the outcome should not have hinged on such a technicality, but on the extent and nature of the surveillance, which can only be tested in terms of a “reasonable expectation of privacy”.

Furthermore, trying to impose an 18th century solution on a 21st century problem is an exercise in futility – it completely fails to address the underlying problem of the rapid encroachment of technology on our privacy. Justice Alito jokingly asked us to imagine a constable hiding in a coach in 1791 (the year the Fourteenth Amendment was ratified) in order to monitor its owner’s movements.

Turning once again to Roy Olmstead, who, after serving four years in prison, was granted a full presidential pardon in 1935, we note that his case spawned one of the most noteworthy dissenting opinions in judicial history. Though heavily criticized at the time, Justice Louis Brandeis’s dissent in Olmstead that eventually became the law:

[The Framers of the Constitution] sought to protect Americans in their beliefs, their thoughts, their emotions and their sensations. They conferred, as against the government, the right to be let alone—the most comprehensive of rights and the right most valued by civilized men. To protect that right, every unjustifiable intrusion by the government upon the privacy of the individual, whatever the means employed, must be deemed a violation of the Fourth Amendment.

Moving forward, we should only hope that, at least the judicial branch of our government never forgets.

Opinions

Banging the Close in the Oil Markets

“Banging the close” is a practice in which traders place a large volume of orders just before the close of an exchange. The practice, often with the aid of high-speed computers and sophisticated algorithms, attempts to influence the settlement price of the day with the accumulated large positions.

In a 50-page complaint filed in federal court in New York in 2008, the CFTC alleged that Optiver, a Dutch high frequency trading firm, engaged in “banging the close” on 19 different occasions in 2007 in connection with crude oil, heating oil, or gasoline contracts. According to the CFTC, the firm affected prices on at least five occasions, and reaped a profit of approximately $1 million. Optiver’s strategy, the CFTC charged, was to execute approximately 20 to 30 percent of the firm’s futures trades just before the close and the remainder during the close. Because the futures contracts traded offset the Trade At Settlement contracts, Optiver would profit if it could trade such an amount of its futures contracts before the close.

Optiver’s colorful language, in recorded telephone conversations and emails, gave the government an easy target. The firm developed what it called the “Hammer,” the aptly-named, secretive trading software that allows its users to rapidly enter orders and even jump to the head of the trading queue, which operates on a “first in, first out” basis. Putting your orders first in line makes it possible for “bullying” and, thus the subtle price manipulation, to work in your favor. But perhaps the most damning pieces of evidence in Optiver’s case were emails and phone recordings involving three Optiver employees (also named as defendants in the complaint) discussing building up positions in order to “bully” the market, referring to the scheme as a “fun game” and, in some cases, even bragging about the profits generated. In one recorded conversation, the CEO of Optiver US told a trader, “…you should milk it for right now because you never know how long it’s going to last.” To make matters worse, Optiver employees made false statements to NYMEX compliance officials in order to conceal their manipulative intent.

Optiver’s antics aside, the question is what really constitutes price manipulation? Traditionally, price manipulation allegations have been exceedingly hard to prove, and most of the cases brought by the CFTC are settled, making it difficult to find widespread agreement on standards of proof. While courts differ considerably on what these standards should be, the Southern District of New York1 tends to more or less use the following elements: (1) the defendant possessed the ability to influence market prices; (2) an artificial price existed; (3) the defendant caused the artificial price; and (4) the defendant specifically intended to cause the artificial price. Whether or not the CFTC could prove these elements about Optiver is debatable.

The CFTC would probably take a broader view, that any scheme such as “banging the close” amounts to manipulation, whatever the outcome. So is an attempted manipulation against the law, like attempted murder? As interesting a question though it may be, that is a topic for another day.

Even without going to trial, though, Optiver’s circumstances painted a noticeably unflattering picture of company in the public eye and, not surprisingly, Optiver agreed to a settlement with the CFTC, which included a $13 million fine and disgorgement of its alleged $1 million profit.

But beyond just the legal implications, the Optiver case has also become an argument in the political arena focused on two issues: the high prices of gasoline and “excessive” market speculation. In fact, just two days before the announcement of the Optiver settlement, President Obama called on Congress to increase civil and criminal penalties on speculators that artificially manipulate markets and drive gasoline prices higher. Regardless of the reasons behind such rhetoric, the argument seems somewhat disingenuous as, in actuality, high gas prices have very little to do with market manipulation and everything to do with oil supply and disruptions, unrest in the Middle East, lack of alternative energy sources and global politics in general.

The other issue that the Optiver case brings to mind is the effort to curb market speculation, especially with respect to the energy markets. The CFTC’s approval of position limits last October is just one example of the government’s intent to accomplish this. But, the truth is that speculation and manipulation are often confused with one another and this perpetuates widespread misunderstanding. Healthy market speculation is vital to the economy and provides short-term and long-term benefits, including accurate pricing information and, more generally, enables risk transfer and better hedging and financial outcomes for producers and consumers.

The benefits of high-frequency trading (HFT) are harder to discern. In a way, HFT is the new Wild West of the financial markets, with regulators struggling to keep up with constantly advancing technology aimed at finding new ways of turning profits. What is clear is that HFT firms like Optiver have attracted some of the brightest minds in finance and, thus, are at the forefront of innovative technology. But, precisely because the technology is so advanced, HFT has naturally also attracted considerable scrutiny and with it increased suspicion, especially with respect to the real function and objectives of such practices. If the motive is a quick profit and the methods turn out to be primarily front running, bullying, scalping and sub-pennying as opposed to genuine speculative market activity (e.g., market making), then regulators will have a much easier time deciding which side of the speculation-manipulation spectrum HFT falls under, and a much greater inclination toward heavier regulation and enforcement.

In re Natural Gas Commodity Litigation, 337 F.Supp.2d 498, 507 (S.D.N.Y. 2004).

Opinions

And Now For Something Completely Different*: When A Hedge Isn’t

The verb form of the term hedge dates back at least to the 16th century. In Shakespeare’s The Merry Wives of Windsor, Sir John Falstaff said, “I, I, I myself sometimes, leaving the fear of God on the left hand and hiding mine honour in my necessity, am fair to shuffle, to hedge and to lurch.” Falstaff was a vain, overweight, drunken liar. So it’s no surprise that he would equivocate.

The term was first used in financial transactions, in the early 17th century, when “hedging one’s debts” referred to a loan secured by inclusion in a larger loan. The more familiar “hedging one’s bets” came into usage somewhat later when bookies of the day laid off a bet by taking out smaller bets with others to avoid going broke to pay off the original larger debt.

All of this came to mind when we took notice of the debate about whether J.P. Morgan’s much publicized trading loss, reported to be somewhere in the billions of dollars, could possibly be, as the bank claimed, attributed to a hedge gone wrong. At the risk of over simplifying it, the trade basically consisted of the bank’s selling a CDS index as a form of insurance which depended on the stability of the corporate bond market.

So the question is, what’s a hedge? And how much does the definition really matter?

In the most general sense, a hedge is an investment undertaken to limit the risk of another investment. But hedging is by no means confined to the world of investments. One can hedge a long position with put options or by buying insurance to limit the risk of injury. Insurance after all is nothing more than put writing in inefficient markets.

Other more technical, and more complicated, definitions govern where the existence of a hedge has regulatory significance. For example, the Commodity Futures Trading Commission exempts hedges from position limits for derivatives. To qualify, a transaction must be economically appropriate for the reduction of risk in a business and satisfy a number of additional requirements.1

And transactions in commodity futures, even if only partially designed to protect against principal risk for a business, are incident to the everyday operation of the business. Such hedges therefore give rise to ordinary income, or ordinary deductions, and are not entitled to capital gains treatment.2 So whether a transaction is a hedge does have practical consequences.

These consequences and the definition of a hedge only confirm what we already know about the nature of hedging. A hedge protects to some degree against a risk of some sort, and you need to know about the underlying facts to determine what is a hedge and what isn’t.

One characteristic of a hedge is its liquidity or at least a limit on the downside. To a large investor with large positions, the problem of maintaining a comparatively liquid hedge is particularly acute.

But back to J.P. Morgan. The bank’s CDS positions are relatively illiquid. It does not yet know the extent of the losses. Still, J.P. Morgan’s claim that the position was a hedge, at least at the outset, is hard to refute. A bank that big has so many profit centers and positions on its book, that any transaction is conceivably a hedge.

Maybe you already know this. But what starts out as a hedge turns into something else‑‑say, a bad investment–when it gets to be bigger than the position it was supposed to be hedging, moves the market and, most importantly of all, winds up causing the investor to lose a lot of money without a more than offsetting profit in the primary position elsewhere.

* Used without permission of Monty Python.

17 C.F.R. §1.3(z).
Corn Products Refining Co. v. Commissioner, 350 U.S. 46 (1955).

Opinions

Intrade No More

At the International Exhibit of 1884 in London, Sir Francis Galton’s curiosity was piqued by a weight-judging competition, in which contestants guessed the weight of an ox to be “butchered and dressed.” About eight hundred people tried their luck at sixpence a ticket. Experts and non-experts alike competed.

Galton, a Victorian polymath, was not so much interested in the competition itself as he was in the collective prediction of the participants. He had set out to prove that there would be no significance to the crowd’s guess, believing as he did that a few dumb people would certainly lead the crowd to a dumb answer. But the result was surprising. On average, the crowd as a whole predicted the ox weighed 1,197 pounds. Butchered and dressed, the ox actually weighed 1,198 pounds. The average was closer than the guess of any individual contestant.

Galton’s impromptu experiment was early support for the view that the aggregation of information for a group results in decisions that are better than the decision of any single member of the group. Today, thanks to technology, this theory is being tested over and over again.

Intrade is an online trading exchange, founded ten years ago in Dublin, that allows individuals to bet on a wide range of outcomes involving future non-sporting events. Intrade participants, taken as a whole, accurately predicted not only the result of the 2012 Presidential Election, but also the results in each state save Florida, which was given a 69% probability of going to Mitt Romney. In fact, Intrade’s predictions of the outcomes of the 2004 and 2008 presidential elections were just as accurate. But how does Intrade really work?

Basically, on the Intrade platform, all events are defined by a “yes” or “no” propositions with a final share price of $10.00 for a “yes” outcome and a final share price of $0.00 for a “no” outcome. The share price fluctuates with the opinion of participants. The Intrade odds on Rick Perry’s winning the Republican nomination fell by 50% during a primary debate only seconds after he forgot the federal department he wanted to eliminate. When you buy shares, you are betting “yes” the event will occur, and when you sell shares, you are betting “no” the event will not occur. The contract may trade in the range of 1-100, where 1 equals $0.10 in value. Because all shares are bought and sold within this range, a participant may either profit or lose at “settlement,” depending on what the share price was when it was bought or sold.

At first glance, Intrade looks like a bookmaker. But the resemblance is superficial. For one thing, Intrade does not charge any transaction fees or commissions. For a flat $4.99 per month, a participant can make an unlimited number of predictions on questions about politics, entertainment, financial indicators, weather, current events and legal affairs. But perhaps more importantly, its prediction capability makes Intrade a valuable information tool. Intrade’s predictions and odds provide governments, academia and the markets with fairly reliable information about the future. So why does the Commodity Futures Trading Commission (CFTC) have a problem with Intrade?

Intrade, the CFTC argued in a complaint filed in a federal court, is taking bets on commodity options (like the prices of gold, crude oil and other things) without listing the contracts on a regulated exchange. Offering commodity option contracts to US customers for trading as well as soliciting, accepting and confirming the execution of orders violate of the Commodity Exchange Act’s ban on off-exchange options trading. Intrade, having been in the CFTC’s crosshairs for several years now, finally decided to shut down its platform to the US market this past November. As of December 23rd, all US Intrade accounts (which comprised approximately 40 percent of all its accounts) have been suspended.

Does Intrade fall within the CFTC’s jurisdiction? A prediction market through which customers buy and sell binary options allowing them to predict whether a future event will happen certainly appears to be within the purview of the Commodity Exchange Act, which regulates a broad range of transactions–futures, swaps, options on futures, swaps and commodities–in connection with commodities, property and events.

But the real question is whether the CFTC should exercise its jurisdiction. A similar prediction market run by the University of Iowa, the Iowa Electronic Markets (IEM), operates pursuant to a CFTC no-action letter. It is largely for academic and research purposes and caps bets at $500, but real bets are being made utilizing IEM every day. Evidently, even the CFTC sees some utility in prediction markets.

Additionally, while some of the betting that Intrade hosts has a connection to traditional commodities (i.e., betting on the future prices of commodities), other topics have no relation to commodities. For example, it is hard to see how a bet on whether Lindsey Lohan will get arrested before June 30 has any real impact on the integrity of our futures markets. In this sense, it seems to us that the CFTC’s actions with respect to Intrade may have been unwise and that the agency probably has better ways to utilize its resources.

Furthermore, given the fact that even the CFTC has recognized the utility of prediction markets, it would seem that there should be some middle ground when it comes to Intrade. One solution could have been for Intrade to simply eliminate its bets on physical commodities, or for the CFTC to regulate only bets on commodities. Regardless, the removal of Intrade from US markets altogether does not, in our view, have sufficient public policy justification. As it stands, with Intrade out of its largest market, its forecasts will undoubtedly have somewhat diminished predictive power.

Galton wrote several years after his experiment, “I know of scarcely anything so apt to impress the imagination as the wonderful form of cosmic order expressed by the law of error. The huger the mob and the greater the anarchy, the more perfect is its sway. Let a large sample of chaotic elements be taken and marshalled in order of their magnitudes, and then, however wildly irregular they appeared, an unexpected and most beautiful form of regularity proves to have been present all along.”

If there ever was a utility for Intrade, it might have been this. And while Intrade may be no more in the US, prediction markets are not going to go away anytime soon.

Opinions

Upcoming SEC Certification Requirements for Officers/Directors of Private Funds and New Quarterly De

In July of 2013, the U.S. Securities and Exchange Commission (“SEC”) adopted a “bad actor” provision to be added to Rule 506 of Regulation D. These amendments are aimed at disqualifying securities offerings involving certain “felons and other bad actors”. The change potentially affects all hedge fund directors and/or any “Covered Person,” as each such person is advised to make a certification of compliance to the SEC. Covered Persons include, among others, directors, executive officers (of funds or fund management companies), beneficial owners of 20 percent or more of an issuer’s voting securities, promoters and investment managers. Non-compliance may have an impact on the ability of funds to attract new investments and, in some cases, may trigger enforcement actions by the SEC. Covered Persons are expected to be in compliance with the new rule by September 23, 2013.

All hedge fund directors are encouraged to certify that they are not disqualified because of:

Criminal convictions (felony or misdemeanor in connection with sale or purchase of a security);
Court injunctions and restraining orders;
Final orders (as defined in Rule 501(g)) of certain state regulators (such as securities, banking, and insurance) and federal regulators, including the U.S. Commodity Futures Trading Commission (the “CFTC”);
Commission disciplinary orders relating to brokers, dealers, municipal securities dealers, investment advisers, and investment companies and their associated persons;
Certain SEC cease and desist orders;
Suspension or expulsion from membership in, or suspension or barring from association with a member of, a securities self-regulatory organization (“SRO”);
Commission stop orders and orders suspending a Regulation A exemption; and
U.S. Postal Service false representation orders.

If you have any questions about these amendments, you should contact Crow & Cushing so that we may assist in helping you come into compliance with the new rule by making the appropriate certifications. Please contact us with any questions that you may have.

For more information on the new Regulation D amendments, please visit: http://www.sec.gov.

With respect to all registered commodity trading advisors (“CTAs”), the National Futures Association (“NFA”) requires all CTAs to submit a Form PR each quarter. The Commodity Futures Trading Commission (“CFTC”) adopted these rules on March 30, 2013. The first filing for NFA Member CTAs will be for the quarter ending September 30, 2013 and will be due on November 14, 2013.

Please note that NFA’s quarterly Form PR consists of the CFTC Annual Form PR along with some additional questions relating to certain trading programs being offered by the CTA, and the related monthly rates of return and the assets under management for those trading programs.

You may view the new Form PR here: http://www.nfa.futures.org/NFA-electronic-filings/CTA_PR_Template_for_Members.pdf