Too Big to Fail: Big Tech Plays the Monopolist’s Get Out of Jail Free Card

Amid the backdrop of a US Copyright Office that all but dismissed Big Tech’s arguments that training generative AI on copyrighted creative works without the consent of the owners of those works is “fair use,” various tech moguls, including, most recently, former Facebook/Meta executive Nick Clegg, are leaning hard into the claim that requiring AI companies to license creative works they want to use as training data would “basically kill the [UK’s but you can insert country here’s] AI industry.” These AI-profiteering hype men then assert that the purported demise of the AI industry that enforcing copyright laws would bring about would also threaten national security. China, we are continuously told, will not obey copyright laws and will, therefore, be able to get ahead and become the dominant global superpower. So powerful and important to national security is this tech that is somehow extremely cost and resource-intensive to train yet easily replicated for significantly less, much less, and, soon after, much, much less and that both makes up data (hallucinates) and leaks out real and sensitive data all while empowering and emboldening con artists, destroying the environment, eliminating the livelihoods of the very people it depends on to exist (artists and other copyright holders), and flooding once-human spaces on the Internet with bots and slop.

I digress. On closer examination, the argument that requiring tech companies to license copyrighted work they use to train generative AI would (a) “kill” the American generative AI industry, which would (b) create a vacuum for China to overtake the United States in the fight to be the dominant global superpower (ruling in harmony and forever with other Global North nations) is a rebranding of the same Too-Big-to-Fail rationalization that American corporations used to justify receiving government handouts during the 2007–2009 Great Recession while everyday people lost everything. Consider the following.

Deregulation and Regulatory Capture: Parallels with Big Tech’s AI Moves

Between 1970 and 2007, the finance, insurance, and real estate (FIRE) industries succeeded in deregulating banks and savings and loans (S&L) institutions, eliminating (among other crucial pieces of legislation) the Glass-Steagall Act, which Mother Jones notes, was created in response to the 1929 stock market crash as a way to separate “‘commercial banks’ focusing on consumer activities (checkings, savings) from ‘investment banks,’ which deal with speculative trades and mergers.”

In 1934, President Franklin D. Roosevelt signed the Securities and Exchange Act to stop the practice of stock buybacks—a form of market manipulation and/or insider trading in which corporations (or their employees) artificially inflate the apparent value of their stocks by buying up shares of their own stock in order to create scarcity and raise prices. Vox reports that the Securities and Exchange Act of 1934 was successful. Instead of engaging in market manipulation, companies mostly (a) reinvested profits into their companies (e.g., “built new factories or created new products”), (b) hired more employees and/or raised wages, or (c) paid investors dividends. According to Vox:

Most American companies did a mix of all three, with the bulk of profits going towards reinvestment and wages. . . . The productivity of American workers nearly doubled in the 30 years after World War II and so did hourly wages. This helped build the American middle class. But things didn’t stay that way. Productivity kept rising, but wages flatlined.

Republican Ronald Reagan became President and significantly eroded the Securities and Exchange Act in 1981 by appointing the Wall Street-friendly John Shad as head of the Securities and Exchange Commission. Shad favored companies giving more profits to investors or shareholders rather than reinvesting them into the companies (e.g., improving products and building factories) or increasing wages, and so he simply changed the rules to allow stock buybacks.

Vox’s video displays a chart that shows the direct correlation between the deregulation of the Securities and Exchange Act of 1934 and the stagnation in wages that followed. Note, however, that one group’s wages did not flatline. CEO pay catapulted since executives often use stock buybacks as a way to inflate their own salaries, which are directly tied to company stock prices. Vox points out that, due in large part to stock buybacks, on average, American CEOs make 220 times more than their employees. The Institute for Policy Studies reports that the disparities are even starker at companies that pay the lowest wages, with CEOs making 603 times more than the average pay of their employees.

Despite this, stock buybacks are now ubiquitous. Alarmingly, Part 1 of Reuters’ special report The Cannibalized Company points out that stock buybacks are sometimes greater than the net income of the companies that engage in them. As a result, Vox notes, American workers are paid much less relative to their productivity. Vox also asserts that American companies, such as General Motors (GM), have lost global market share by prioritizing stock buybacks at the expense of reinvesting in their companies (in research and development, for example) and in workers, who are often behind new designs and innovations that align with changing consumer preferences.

As Reuters observes, instead of investing internally in research and development, American companies now buy or invest in startups (e.g., Google’s acquisition of DeepMind and Microsoft’s part ownership of OpenAI). However, one wonders what impact this has on innovation once these startups are absorbed into companies that prioritize short-term profits over workers and long-term, stable growth.

Also concerning is the extent to which stock buybacks enable company executives to manipulate markets, inflating prices for their stocks even when, as in the case of Hewlett-Packard (HP), a company is underperforming or, as in the case of Boeing, a company is prioritizing investor and executive pay at the expense of public safety. Nonetheless, the stock buybacks have continued unabated, hitting their highest level since 2018 in 2024.

Some other key points Mother Jones highlights in America’s tumble toward the Great Recession include the following:

• Beginning in 1970, financial interests began successfully chipping away at regulations like the Glass-Steagall Act.
• In 1982, the Garn-St. Germain Depository Institutions Act deregulated the savings and loan (S&L) industry.
• In 1996, the Office of Thrift Supervision—much as Trump’s One Big Beautiful Bill Act threatens to do for generative AI—“issue[d] a rule preempting almost all state laws regulating S&L credit activities.”
• In 1999, after the finance, insurance, and real estate (FIRE) industry funneled massive donations to elected officials, including $4.6 million to Texas Republican Senator Phil Gramm, “the Gramm-Leach-Bliley Act gut[ted] Glass-Steagall, setting off a wave of megamergers among banks and insurance and securities companies.”
• On June 20, 2000, “Treasure and HUD urged Fed to investigate subprime units of major banks. No Fed action follows.”
• In December 2000, “as Congress head[ed] for Christmas recess,” Senator Phil Gramm snuck in “a 262-page amendment to an omnibus appropriations bill.” This Commodity Modernization Act, writes Mother Jones, “deregulate[d] derivatives trading, giving rise to [the] Enron debacle, and open[ed the] door to an explosion in new, unregulated securities.” (original emphasis)
• In March 2001, the “FTC sues Citigroup and its subsidiary associates . . . charging ‘systematic abusive lending practices’ . . . 18 months later Citigroup settles for a paltry $215 million.”
• Between January 2001 and January 2007, various statewide laws that barred predatory lending were quashed by tech-centric subprime lender Ameriquest. Ameriquest employees donated $200,000 to George W. Bush’s campaign, and founder Roland Arnall, whom the Los Angeles Times notes Bush would later make ambassador to the Netherlands even as Ameriquest was settling a $325 million lawsuit for “allegations that it deceived borrowers, falsified loan documents and pressured appraisers to overstate home values,” donated (according to Mother Jones) “more than $5 million to pro-Bush PACS.” Another Los Angeles Times article reports that “Ameriquest blamed a computer glitch for the alleged violations.” Expect many more similar “computer glitches” facilitated by AI-centric companies in the future.
• Between January 2001 and January 2007, the FIRE sectors engaged in regulatory capture through the “Responsible Lending Act,” which Mother Jones asserts was “designed to preempt stronger state laws” and which “consumer advocates [called the] ‘Loan Shark Protection Act.’” They also succeeded in making it harder for consumers to sue them.

This brings us to the beginnings of the 2007–2009 financial crisis, aka the Great Recession.

Failing Big Two: Repeating the Mistakes of the Great Recession

Senator Elizabeth Warren (D-Mass) has made the convincing argument that Glass-Steagall and other regulations that the FIRE sectors weakened or repealed had drastically curbed the every-15-years, bank-boom-and-bust cycle that characterized the US economy in centuries prior. In an interview with CNBC Ambition, Warren underscored that, thanks, in large part, to Glass-Steagall, the United States avoided a major recession for more than 50 years.

According to the History Channel, when the housing market began to boom, post-deregulation, in the early 2000s, banks and other lenders gave loans to people with bad credit whom they knew were unlikely to be able to pay back those loans. Over the next few years, the real estate market became saturated with underqualified borrowers who had speculated that property values would rise, enabling them to make a profit, and lenders who also speculated that they could profit from charging these borrowers high interest rates and, when they eventually failed to make payments on their loans, could profit further by seizing the properties, which (if the trend continued) would have appreciated in value.

In 2007, however, the housing-market bubble burst, resulting in millions of foreclosures and leaving financial institutions with properties that were worth far less than they had anticipated. Both subprime mortgage borrowers and lenders lost money on the bad loans they had gotten or made, causing a downturn that spread to the rest of the economy: a Great Recession facilitated primarily by this “subprime mortgage lending crisis,” as well as by the deregulation of financial institutions that had preceded it and the excessive borrowing that this deregulation had also enabled.

Nonetheless, banks had been emboldened by decades of deregulation and corrupt liaisons with public officials. They and elected officials with financial ties to them claimed that these banks and businesses deserved exceptional treatment, that they were “too big to fail”—which Marketplace explains is the idea that the systemic risks posed by allowing certain private-sector corporations or institutions to collapse, even when they have behaved recklessly, are too great and, therefore, governments (in other words, taxpayers) must bail them out when they are at risk of collapse.

The big banks borrowed from a familiar playbook to get what they wanted. Like Ameriquest, which Mother Jones reports went out of business toward the beginning of the crash in August of 2007 and which had intimidated states into caving in to its deregulation demands by refusing to lend in and getting other financial institutions to stop lending in these states when they attempted to fight predatory lending practices, major banks froze credit to small businesses and consumers when the federal government initially hesitated to bail them out, precipitating a nationwide recession. In 2008, even as millions of everyday Americans, including those who had not been part of the speculative real estate market, lost jobs, businesses, incomes, and homes (a precursor to the AI-industry-manufactured crisis artists and other workers face today)—the federal government bailed out the very same banks that had caused the crisis, giving them government handouts (taxpayer money) to the tune of $700 billion through the Troubled Asset Relief Program (TARP) and cementing the problem of moral hazard in the American financial sector.

How did these financial institutions use those taxpayer-funded government handouts? According to Reuters’ Anna Yukhananov and Forbes’ Steve Schaefer, a report issued by the special inspector general who investigated TARP found that the US Department of Treasury’s Office of the Special Master—“pay czars” Kenneth Feinberg and, later, Patricia Geoghegan—approved payments to CEOs of banks that received TARP funds well in excess of $500,000 even though Feinberg had “‘testified before Congress that “base salaries should rarely exceed $500,000 . . . and should be, in many cases, well under $500,000.’” More than 54% of executives at companies that received TARP funds got more than $3 million in executive pay and only one of the 69 executives received less than $1 million.

How did this happen? The inspector general noted that, “Without developing some criteria . . . Treasury put itself in a position of essentially letting companies drive what pay Treasury was approving.” This is what Trump’s Big Beautiful Bill, which would prohibit state-level AI regulation (including eliminating AI-related laws states have already passed) for 10 years, would also do for AI companies.

Faced with a Republican-led federal government that seems committed to never regulating generative AI yet seeks to quash state-level regulations, under Trump’s Big Beautiful Bill, AI companies will be free to build and maintain technology that helps employers and other powerful decision makers discriminate based on protected characteristics, such as race, gender, and age; violates privacy on multiple levels; enables criminals; drastically accelerates climate change; and causes other harms—all while AI companies continue to pillage artists’ works without our consent and without crediting and compensating us. The AI clause in the Big Beautiful Bill is, essentially, a government handout to Big Tech companies—to those who least need it—at the expense of artists, one of the most financially at-risk group of workers in the country, and other vulnerable groups, including women, older Americans, and racial minorities.

“National Security” and “But China!”: The New Too Big to Fail for the AI Era

As Warren Buffett noted in a Wall Street Journal interview in which he reflected on TARP and, more broadly, the 2007–2009 Great Recession:

I can understand why people that lost their houses, lost their jobs . . . feel that there must be somebody out there that was profiting from this that did it doing some things that should send them to jail. The people that ran most of the institutions, the big institutions that got in trouble, . . . they went away rich. They may have been disgraced to some degree, but they went away rich, so I don’t think the incentive system has been improved a lot from what it was [in the time leading up to the crisis]. (emphasis added)

Today, we are being fed the same trickle-down tall tale. Artists and others whose copyrighted works have been ingested without their consent to train machines intended to replace them are expected to believe that today’s corporate robber barons are actually do-gooders in disguise whose goodwill will, at some point after they are finished monopolizing creative markets and have laid off many more workers, eventually benefit the very people they’ve exploited, maintaining US dominance over the current world order and setting us on the path to a techno-utopia in which no one, not even today’s laid-off workers, will even need a job . . . if we would just roll over and allow AI companies to operate with no regulation or accept the kind of deregulation or regulatory capture that Trump’s so-called “Big Beautiful Bill” aims to facilitate.

Tech corporations, borrowing from Ameriquests’ and banks’ freeze-credit-to-force-deregulation playbook, also threaten to withhold services from countries that attempt to regulate AI. Again, we’ve been here before.

Oddly, many continue to fall for these tired tactics even when the value of AI is so fragile relative to the value of the stolen intellectual property that undergirds it that anyone and everyone could simply do what China’s DeepSeek has done: steal from the thieves. The fact that American AI companies and their cronies have responded to the rise of DeepSeek by pushing for the United States government to protect the intellectual property of AI companies without protecting the artists these American companies steal intellectual property from is a level of irony so rarely encountered that one could call it blatant hypocrisy and, in fact, many do.

As I wrote previously, no thief wants to be stolen from. DeepSeek’s apparent distillation of ChatGPT underscores the fact that the real value of generative AI is in the training data—the intellectual property created by artists and other copyright owners. Training data is the only variable in the otherwise constant AI equation that differentiates models. AI models are plastic shells; the training data (copyrighted works) are the authentic pearls hidden within them. Wise opportunists with a similar lack of moral standards will simply wait for American counterparts to waste money on building expensive models, distill these models, and then fine tune them on better data. (Yet most countries are intent on protecting the plastic shells, AI models, and not the pearls, the creators of the data that allow these models to function well.) AI companies with exclusive access to high-value data will, therefore, almost certainly beat out those with limited or no access to it.

Perhaps in recognition of this reality, consider that, despite Clegg’s claims, numerous generative AI companies that license the work they use to train generative AI already exist. See FairlyTrained.org for examples. Moreover, as the Centre for Regulation of the Creative Economy (CREATe) shows, most, if not all, of the largest Big Tech companies have licensed or are licensing content from major rights holders. The Financial Times’ Melissa Heikkilä reports that “fledgling groups [that help AI companies license high-quality training data] such as Pip Labs, Vermillio, Created by Humans, ProRata, Narrativ and Human Native” are also popping up. This is why the US Copyright Office concluded:

Voluntary licensing is already happening in some sectors, and it appears reasonable or likely to be developed in others—at least for certain types of works, training, and models. Where licensing options exist or are likely to be feasible, this consideration will disfavor fair use under the fourth factor.

Also:

In our view, American leadership in the AI space would best be furthered by supporting both of these world-class industries that contribute so much to our economic and cultural advancement. Effective licensing options can ensure that innovation continues to advance without undermining intellectual property rights. These groundbreaking technologies should benefit both the innovators who design them and the creators whose content fuels them, as well as the general public. (emphasis added)

Nonetheless, just as they did during the Great Recession, major corporations want communism for capitalists. Tech companies seek to legalize theft from the most vulnerable, such as independent artists, while working out deals and settling lawsuits with those who can afford to fight them on equal footing—that is, major rights holders. They seek to both socialize the risks of generative AI through widespread intellectual property theft and to privatize the profits.

There is no less risky business model than one built on legalized theft, and in the most unjust societies, it is those with power who define what is legal. All those who have profited from unpaid labor across the centuries understood this.

Today, American fintech seeks to execute a modernized take on the repressive monarchies and fledgling democracies that represented the interests of elite white men—to move us forward, they say, by pushing us backward. They seek to override the will of the people that overwhelmingly supports artists on the issue of generative AI at, according to Variety’s “Generative AI & Licensing: A Special Report, rates of 77% and 75%, respectively, on questions of consent and compensation, and 76% on the question of whether “there should be restrictions on the ability of AI tools to copy the style of a specific creative artist/writer/musician/creator.” American fintech also seeks to override democratic institutions like the US Copyright Office and state legislatures through brazen corruption: buying elected officials at the federal level.

But the times have changed. When America was a fledgling democracy and the British monarchy had more control over the UK, both countries’ populations were far less educated and had fewer options to fight subjugation at the hands of despots who terrorized their subjects into accepting that they were divinely ordained to rule over them. When will the tech sector—the industry that most seeks to shape the future and which is teeming with people who believe their vision of what the future should look like supersedes everyone else’s—realize this?

A note: Artists Resisting Exploitation (ARE) will be on an indefinite hiatus. But if you want to read more of ARE’s writing, you can view a short post I made on Reddit last month about Part 3 of the US Copyright Office’s “Copyright and Artificial Intelligence” reports, as well as the subsequent exchange with some folks in the opposition, here.

Thanks to those who have supported ARE, and thank you, especially, to paid subscribers. Whether you were able to pay for just a few months or more or couldn’t afford to become a paid subscriber but genuinely support ARE’s work, I’m immensely grateful for your solidarity. If you currently have a paid subscription, I've paused it so that you are not charged during the hiatus and won't be charged, without your express permission, in the future. If you are a newbie who is just now coming across ARE and want to support this work (to make it possible for ARE to keep existing), please use the Donate button below. Thank you!