This image is a section of the Detroit Industry Murals by Diego Rivera, created between 1932 and 1933. Rivera, a renowned Mexican muralist, painted these panels inside the Detroit Institute of Arts. They are widely considered one of his greatest works and a landmark of 20th-century art.

The workers shown are laboring on automobile assembly lines, likely at the Ford Motor Company’s River Rouge Plant.

Rivera celebrated the dignity and effort of industrial labor, portraying the workers with strength and unity.

The style is Social Realism, characterized by its focus on working-class subjects and its detailed, lifelike forms.

Photograph of a detail of a 1933 Diego Rivera fresco at the Detroit Institute of Arts. Justin Hendrix/Tech Policy Press

A Labor Day Prescription for Preserving Fair Employment in the Digital Age

A large number of people take part in a gathering of boats and motorbikes in support of the people of Ceuta on 30 August 2026 in Ceuta (Spain). (Europa Press via AP)

A large number of people took part in a gathering of boats and motorbikes in support of the people of Ceuta on 30 August 2026 in Ceuta (Spain). (Europa Press via AP)

How Disinformation Fueled the Latest Ceuta Mass Crossing

Mechanical Turk is Closing. The Workers Who Built AI Are Still Here

Members of California's legal team leave court after a settlement was reached in a multi-state lawsuit accusing Meta of causing harm to children on social media outside the Ronald V. Dellums Federal Building and US Courthouse on Wednesday, Aug. 26, 2026, in Oakland, Calif. (AP Photo/Noah Berger)

August 2026 US Tech Policy Roundup

The Colorado State Capitol is framed by buildings along Sherman Street Friday, June 27, 2025, in Denver. (AP Photo/David Zalubowski)

What 12 State ‘Companion Bot’ Laws Demand of AI Providers

A tic-tac-toe board with human faces as digital blocks, symbolizing how AI works on pre-existing, biased online data for information processing and decision-making.

Amritha R Warrier & AI4Media / Better Images of AI / tic tac toe / CC-BY 4.0

Germany's Draft Law on AI in Migration Raises Rights and Bias Concerns

A collage of circuit boards interwoven with aerial photographs of water treatment facilities, agricultural lands, industrial sites and mining machinery.

Privacy by Architecture Makes AI Compliance Work

A yellow school bus is seen in Chicago, Illinois, United States, on July 28, 2026. (Photo by Marcin Golba/NurPhoto)

In the US, Caution Rules on AI Ahead of K-12 School Year

A young child holds up a smart phone in front of their face using both hands.

This Week's Meta Settlement Will Reverberate Across the World for Years

This image is a collage with a colourful Japanese vintage landscape showing a mountain, hills, flowers and other plants and a small stream. There are 3 large black data servers placed in the bottom half of the image, with a cloud of black smoke emitting from them, partly obscuring the scenery.

Deborah Lupton, "Servers in a Landscape" / 

'Technology Freedom Cooperative' To Build Infrastructure for Human Rights and Justice Groups

Meta CEO Mark Zuckerberg (C) ahead of testifying before the Senate Judiciary Committee hearing on protecting children from sexual exploitation online, in the Dirksen Senate Office Building in Washington, DC, USA, 31 January 2024. (Photo by Ritzau Scanpix/Sipa USA)(Sipa via AP Images)

Meta Owes the Rest of the World Its Teen Protections

A massive sculpture of Icarus lies in front of the Concordia Temple in the Valley of Temples in Agrigento, Sicily, on January 5, 2019. Shutterstock

A massive sculpture of Icarus lies in front of the Concordia Temple in the Valley of Temples in Agrigento, Sicily, on January 5, 2019. 

Make AI Companies Criminally Liable for Preventable Harm

The EU and US flags in front of a European Commission logo.

The EU and US Can’t Agree on Tech — or Trust

Nicklas Akers, California Senior Assistant Attorney General, left, and Deputy Attorney General Megan O'Neill exit court after tech giant Meta reached a settlement to end a landmark trial over teen social media addiction in Oakland, Calif., Wednesday, Aug. 26, 2026. (AP Photo/Noah Berger)

Assessing the Landmark Meta Settlement

Signs for company Cambridge Analytica in the lobby of the building in which they were based on March 21, 2018 in London, England. (Photo by Chris J Ratcliffe/Getty Images)

The $459 Million Cambridge Analytica Reckoning Buried in Meta's Teen Safety Settlement

The TikTok logo is displayed through a magnifying glass in this photo illustration in Ontario, Canada, on February 5, 2026. (Photo by Thomas Fuller/NurPhoto via AP)

Understanding Brazil’s $29.7 Million TikTok Fine Over Children’s Data

Three human silhouettes are depicted mid-stride, each comprising of a collage of hardware used to develop AI including circuit boards and CPU chips. In the background there is a black and grey gradient with smoke. Beneath the runners from the left, there are two rectangle panels which run across the image, one is red and the other is blue. 

To Avoid a 'Tokenocracy,' Ensure Popular Access to AI Systems

CHONGQING, CHINA - AUGUST 10: In this photo illustration, the logo of Meta is displayed on a smartphone screen on August 10, 2026 in Chongqing, China. (Photo by VCG/VCG via AP )

What Meta’s US Settlement on Child Safety Means for Europe

Meta CEO Mark Zuckerberg attends UFC Freedom 250 on the South Lawn of the White House, Sunday, June 14, 2026, in Washington. (AP Photo/Alex Brandon)

What's in Meta's Teen Social Media Settlement—and What Hinges on Its Rivals

Image of Solana Larsen with text, Managing Editor Solana Larsen

Solana Larsen Named Managing Editor of Tech Policy Press

Petra Molnar and Liz Carolan are both fellows at Tech Policy Press.

Over two days in late July, an estimated 

 crossed the border from Morocco to Ceuta, a Spanish enclave in North Africa. Many swam a 

 of open sea to make the crossing, and at least 100 

 doing so, according to the enclave’s mayor. In the lead-up to those chaotic days, a wave of disinformation on social media 

 with young people in Morocco that a recent Spanish Supreme Court ruling had opened the border to Ceuta and onto Europe, and that crossing into Ceuta would be easy.

This week, Spain's Prime Minister Pedro Sánchez 

 of the disinformation that preceded the Ceuta crossings, citing research by the European External Action Service, the EU’s diplomatic service, that traced the campaign to Russian and Israeli networks operating alongside what he described as an ultra-right international network that "always uses migration to attack both Spain and Europe." Sánchez said the two governments had seized on the issue because of Madrid's positions on the wars in Ukraine and Gaza, and that investigators had so far found no evidence of Moroccan state involvement. Israel's Foreign Minister Gideon Sa'ar 

, accusing Sánchez of "lying shamelessly," and Russia has also denied any role.

This disinformation effort has striking similarities to patterns seen in 

content targeting people on the move trying to reach the United States

, and in both cases the campaigns worked not because the people who acted on them were credulous, but because the rules governing who may cross a border have been made so opaque, so changeable, and so badly communicated that a video posted by a stranger is often the most accessible account available.

, found dozens of burner profiles on TikTok and Instagram that purported to be young Moroccan content creators who had made the journey and were celebrating a newfound freedom in Ceuta. The videos, structured as influencer content, often contained practical advice on making the crossing, from the equipment needed to advice for children and non-swimmers. While many accounts were shut down, they later reappeared under new handles.

 suggests that young people traveled from across Morocco in response to online rumors and stories, like waitress Dounia Kherradi, who lived a nine-and-a-half-hour drive away, and 16-year-old Ayoub El Abyad, who told reporters that he and his cousin decided to drive 12 hours to the crossing after seeing videos on Instagram. What they found was drastically different than the upbeat soundtracks of Reels suggested; those who had crossed the border told reporters 

 that they had to pass dead bodies on the treacherous swim to Ceuta, that they were met with violence when they arrived, and denied food in stores. They then learned there was no way for them to travel on to Europe and that the rumors had been lies. By the end of the weekend, all but 

None of the mechanics on display at Ceuta are new, and there are clear similarities in the grammar of videos shared of how people on the move try to reach the United States. In March of last year the Associated Press reviewed 

hundreds of TikTok videos posted by Mexican migration facilitation networks

 and found similar themes that El Español documented on the Moroccan side of the Strait; upbeat soundtracks, emoji-laden captions, dune buggies racing toward the border wall, jet skis and helicopters and tunnels offered as premium options, facilitators slicing open cacti so their clients can drink, a field of lettuce captioned "The American fields are ready!,” and, at the end, the satisfied-customer testimonial.

In Ceuta as in the Americas, these videos and imagery function as advertisements, painting a picture of border crossing as a manageable logistical problem with a happy ending.

The commercial layer around this content is by now well established. In Spain, the Prime Minister initially blamed human-trafficking 

 for the disinformation, as they stood to benefit financially from attempts to cross the border.

Meta earned revenue from smuggling and scam advertisements

 targeting people trying to reach the US, and the pattern intensified once the American asylum system was funneled through a single mobile application. When CBP One became the central route to an appointment at a port of entry, a secondary economy grew up around it involving 

 sold to people who arrived at the border to find nothing waiting for them, paid intermediaries charging anywhere from a few dollars to several thousand, and 

GPS-spoofing workarounds circulated in WhatsApp groups

 by people locked out of the app's geographic requirements. The Tech Transparency Project, mapping the same terrain, found Facebook and WhatsApp 

saturated with false claims about policy changes

 and invented special provisions for pregnant women and parents traveling with children.

The rumor that circulated in Morocco in July —that a court ruling had quietly opened a route — is a recognizable instance of a well-documented pattern, and the difficulty it presents is one the coverage of Ceuta has largely avoided. Deceptive content does not reach people on the move through a separate channel from accurate content; it reaches them through the same feeds, in the same visual idiom, and often from accounts that are indistinguishable at a glance.

For example, alongside the burner accounts are the Venezuelan and Colombian creators who 

filmed their own crossings of the Darién Gap

, some of whom accumulated tens of millions of views, and whose videos 

 to shelter locations, transport costs, river crossings and where along the route it was possible to receive a money transfer. Several of them were explicit that they were warning people off rather than recruiting them. The hashtag was viewed 

 This is the same platform and format and for many people it is the only source of information about the journey that is available in a language they read, at a price they can pay, from someone whose position resembles their own.

The framing of the young Moroccans as simply “duped,” therefore, becomes harder to sustain. Scholarship on refugee smartphone use has a precise term for the condition they were operating in. Scholars such as Marie Gillespie, Souad Osseiran and Margie Cheesman describe 

, a state of chronic informational instability in which people on the move rely on peer networks precisely because official channels are inaccessible, untranslated, contradictory or actively hostile, and are exposed to rumor and disinformation for exactly the same reason.

The informational vulnerability is a product of a legal architecture that offers almost no authoritative, usable answer to the question of who may enter and how. In Ceuta, Dounia Kherradi drove nine and a half hours on the basis of what she had seen online because there was no other place to look. Agency exercised inside informational scarcity is still agency, and it is worth being careful that any critique of disinformation campaigns does not reproduce the older colonial habit of locating the problem in the so-called gullibility of people from the Majority World rather than in the systems that keep them uninformed by design.

Moreover, states are among the most active participants in this same information environment. In 2025, the US Department of Homeland Security budgeted 

 for a two-year domestic and international advertising campaign, built around Secretary Kristi Noem telling viewers that "we will hunt you down,” and distributed through 

hyper-targeted social media placements and text messaging

When the spots ran during prime-time football coverage on Mexican television, President Claudia Sheinbaum 

moved to ban foreign government propaganda outright

, reviving a clause repealed in 2014. Deterrence messaging and smuggler advertising run on the same infrastructure, target the same audiences with the same behavioral tooling, and make competing promises about what will happen at the border.

The thousands of Moroccans in July were the target of the same kind of campaign that, in the Americas, is funded to the tune of two hundred million dollars. Neither of these campaigns was an account of what the law actually permitted, and both were persuasion, bought and targeted by parties with a political interest in the outcome. Nobody, in Rabat or Madrid or Washington, was under any obligation to tell people authoritative, timely, and accessible information, and so nobody did.

Labor Day 2026 comes amid rapid adoption of AI systems in American workplaces, with many workers now applying for jobs through AI screeners or finding their performance managed and judged by bossware. At the same time, the Trump administration 

 a foundational civil rights protection that helps ensure employment screens actually reflect what the job requires. While AI technologies undoubtedly offer benefits, without guardrails, these systems can reject qualified people for reasons no one—including the employer—can adequately explain. And because these automated tools are designed using data that does not sufficiently reflect all workers, they can foreclose opportunities for qualified workers based on factors like race, sex, ethnicity, or disability.

We can’t let technological progress turn back the clock on civil rights. To give workers a fair shot in an AI-driven economy, the use of AI must reflect a national commitment to workplace equality. When I served as Chair of the Equal Employment Opportunity Commission, the agency had begun applying existing law to help ensure accountability when the use of technology violated workers’ civil rights. While the landmark 1964 Civil Rights Act was written before anyone imagined a résumé being scored by a machine, the question it asks of any screening device is evergreen: does it actually measure what the job requires? Until Congress enacts new laws to fully address all AI-related risks, traditional civil rights law remains one of the few existing tools to ensure accountability when automated employment systems violate workers’ rights.

Fortunately, the Civil Rights Act’s little-known, but crucial, disparate impact provisions protect qualified workers of all backgrounds from employment practices, including the use of AI, that create unfair barriers based on race, color, religion, sex, or national origin. This means that employers can’t impose unnecessary and irrelevant requirements that disadvantage workers because of a protected trait. As a unanimous Supreme Court explained in 

, such practices discriminate because they operate as “built-in headwinds” to opportunity that are “unrelated to measuring job capability.” Reaffirmed by overwhelming majorities of both the House and Senate in 1991, the provision has led employers to re-examine workplace mandates like arbitrary height and weight restrictions or unnecessary degree requirements and replace them with criteria that reflect the actual knowledge, skills, and abilities needed for a particular job. That’s a win for everyone, because evaluating employees using fair, job-relevant, and merit-based criteria makes for a more qualified workforce.

Outside the AI context, the law against unfair disparate impact has proved invaluable. In 2021, the EEOC 

 to secure relief for women who had been denied employment as truck drivers based solely on a test for which the company had no evidence it was related to the job. The women were qualified—indeed many had done the work successfully in the past—but were rejected by an assessment that did not measure job-relevant skills. Although this victory came in a case initially filed in Trump’s first term, the current administration has attacked this 

—threatening more than six decades of progress toward equal employment opportunity. This endangers the ability to challenge unfair employment practices of all kinds, but the implications are particularly far-reaching for the growing use of AI in employment decisions.

The attack threatens one of the few legal incentives for ensuring that automated employment systems actually work as advertised. That’s because civil rights compliance requires that AI used in employment be job-related and serve a legitimate business purpose. Ensuring that AI used in employment assesses job-related skills will benefit workers and employers—making the process fair and accurate for everyone by evaluating workers on what actually matters to the job. Given extensive evidence that these systems can hallucinate or produce authoritative-looking results based on irrelevant data, this is essential for both fairness and return on investment.

 Justice Department opinion claimed—incorrectly and without evidence—that the law against unjustified disparate impact somehow forces employers to favor some groups over others. Federal law protects all workers, equally prohibiting employment discrimination, including unjustified disparate impact, against individuals of any race, religion, color, sex, or ethnic group. The law also forbids only 

disparate impact, allowing employers to evaluate workers using requirements that are truly job-related when there is no less discriminatory alternative.

The administration also erroneously relies on a recent voting rights case, 

to support its effort to upend disparate impact law. The Supreme Court’s decision in 

 involved intentional, race-based redistricting to create a majority-minority electoral district—a context legally and factually distinct from the Civil Rights Act’s race-neutral employment provisions.

While the law against unjustified disparate impact is not a complete answer to discriminatory or otherwise poorly designed AI, it remains a powerful incentive for employers and AI developers to focus on fairness and to double-check conclusions that a particular AI model accurately predicts job performance. Given the intense economic pressures fueling the current market for AI in employment, such incentives are essential to protect workers and ensure job-related decisions. Ignoring workers’ civil rights as we develop emerging technologies means chasing short-term profits at the expense of long-term gains. Tools that cannot be shown to predict job performance screen out capable people quietly and at scale—at the expense of workers, employers, and the long-term future of responsible innovation.

The law against unjustified disparate impact has brought America closer to our shared value of equal employment opportunity. Rejecting the baseless efforts to narrow it will be crucial to America’s future moral, economic, and technological success.

Twelve states now have passed laws aimed at reducing potential risks associated with “companion chatbots,” particularly for minors. Although the terminology for these products varies, these “companion bot laws” are already in effect in 

The 12 companion bot laws share several common regulatory themes across jurisdictions. Lawmakers have focused on disclosures regarding AI interactions, safeguards for minor users, crisis-intervention protocols, and prohibiting misrepresentations relating to licensed professional services. The laws are not substantively identical, however, with significant differences in both product coverage and the specific obligations or prohibitions applicable to bot providers in a given state.

All of the companion bot laws regulate AI systems that engage users through ongoing conversational interactions. Most of them restrict coverage to bots that exhibit anthropomorphic features and sustain relationships across multiple interactions. The laws in Colorado, Connecticut, Idaho, Iowa, and Nebraska have broader language that cover any bot that “simulates human conversation and interaction” through text, audio or video. With the exception of Hawaii, the laws all provide exceptions for chatbots used for business purposes or for other, narrow functions, such as a limited video game feature.

An analysis of the substantive provisions of the 12 state companion bot laws reveals four overarching principles that state legislatures are effectively requiring chatbot providers to adopt:

1. Transparency and disclosure requirements

All 12 laws require providers to clearly and conspicuously disclose to users that they are communicating with an AI system rather than a human being.

The scope of this requirement varies. Some states require disclosure at the outset of an interaction, while others require repeated or persistent notices during extended conversations. Several statutes impose stricter disclosure requirements when users are minors. Washington’s law goes furthest by requiring providers to prevent chatbots from expressly representing themselves as human to

 user, whereas this requirement exists in several other states only for minor users.

With the exception of New York and Rhode Island, the companion bot laws impose heightened compliance obligations when the user is a minor. While the specifics vary, these obligations generally include “reasonable measures” for

Preventing manipulative or maximizing engagement (

Preventing emotional manipulation and dependency (

, discouraging users from seeking help from adults)

, requirements providers give parents or guardians tools to manage privacy settings and screen time)

Most companion bot laws apply these requirements when a provider knows, or has reason to know, that a user is a minor. Colorado and Georgia go further by imposing an affirmative duty to use "commercially reasonable” methods to estimate user age. However, the Georgia law limits this duty to when it is proportionate to the risk of access to features that may generate sexually explicit content.

, Connecticut adopts a notably different approach. Rather than focusing primarily on reasonable measures to prevent certain outputs, it forbids offering a companion bot to minors if “reasonably foreseeable” that the bot “is capable of” specified harmful interactions. The list of such interactions is broader than the other state laws.

3. Crisis detection and mental health safeguards

All 12 laws require operators to implement protocols that detect user expressions of suicidal ideation or self-harm and refer them to appropriate crisis or mental health resources. Some of the laws require posting of these protocols on the operator’s website. Unlike the other substantive provisions of these laws, these protocols are usually framed as being required for providers as a precondition for offering chatbots in the state. Several of the laws expand the scope of the required protocols to include additional harms such severe emotional crises, violence to others, and eating disorders.

The Colorado, Georgia, and Oregon laws also require the protocols to include escalation procedures for users with repeated or severe crisis indicators. Oregon’s law specifies that providers must use "clinical best practices and expertise" to establish how the AI provides "additional intervention" for a user who continues to express suicidal or self-harm ideation even after being given crisis resources. The only other law with even an indirect reference to clinical practices is Hawaii’s, which requires "evidence-based methods" for measuring suicidal ideation.

4. Restrictions on mental health and professional representations

Several of the companion bot laws prohibit providers from programming or permitting their AI products to represent that they are mental health professionals or can provide professional or licensed mental or behavioral health care. Such restrictions are also found in at least 8 other state laws that focus solely on bots designed for, or representing themselves as capable of providing, mental health. The first three of these laws—in Nevada, Utah, and Illinois—were discussed in a prior DLA Piper 

. Other states following suit are Delaware, Maine, Tennessee, Rhode Island, and Vermont.

 AI law also describes a unique provision that prohibits offering a companion bot to a minor if reasonably foreseeable that the bot is capable of “offering mental health services,” outside tightly defined exceptions.

Colorado’s law extends the concept further by restricting representations related to additional licensed professions, including health care providers, dietitians, and attorneys. These provisions reflect growing concern that users may place undue trust in AI systems that appear to provide professional advice.

Colorado’s proposed rules may shape enforcement

On August 11, the Colorado Attorney General released 

 relating in part to its companion bot law, now called the Chatbot Safety Act. The law does not mandate rulemaking, but the Attorney General indicated that rules would help to clarify compliance obligations. Of particular note, the proposed rules address the statute’s standards requiring providers to adopt “technically feasible measures” and "reasonable measures" to prevent certain categories of outputs to minor users.

For covered products, the law requires providers to institute (a) “technically feasible measures” to prevent specified types of explicit sexual content and (b) “reasonable measures” to prevent “formulating, structuring, or optimizing a response that simulates emotional dependence or isolation from real-world supports.” Sections 11.3 and 11.4 of the proposed rules detail several factors that the Attorney General would consider in determining whether providers had complied with those requirements, including the availability and effectiveness of safeguards and the extent to which the provider engaged in effective testing, monitoring, remediation, and related documentation.

Many of these factors mirror crucial aspects of recommended corporate AI governance programs, including testing, monitoring, and assessing AI products and tools to ensure that they are working as intended and not causing harm. The provisions also indicate the importance of documenting those efforts and responding appropriately to any problematic results.

Given the absence of comparable interpretive guidance in the other states with companion bot laws, Colorado's final rules could become an influential reference point for regulators and courts analyzing similar statutory language. Public comments on the proposed rules—which also cover the recently enacted state law on automated decision-making technology—are due by October 26.

Companion bot regulation is evolving rapidly. In addition to new legislation, state attorneys general and private plaintiffs continue to file cases in which chatbot usage has allegedly caused self-harm, violence, and emotional dependency. Congress is considering multiple AI and chatbot-related proposals, while industry organizations and nonprofits—such as the Better Business Bureau, the Partnership on AI, and Common Sense Media—are developing voluntary standards intended to complement legislative efforts.

The only thing certain in this area is change, with more laws, cases, and other activity on the horizon.

Mechanical gears in black and white photo.

 that Amazon Mechanical Turk (AMT) will close on Sept. 30. It’s a virtual ‘workplace’ that once had more than 500,000 workers signed up.But this isn’t just a story about the end of an aging technology platform. It’s a story of what happens when tech giants are allowed to build businesses based on the exploitation of workers, and then just walk away when the model no longer serves them.

 as part of Amazon’s growing lineup of services. Simply put, it was designed as a marketplace where employers, known as “requesters,” could post work for approved workers to complete. The tasks ranged from surveys and data collection to AI training, content moderation, and government work. They were often referred to as “microtasks” or “data work.” And while a handful of institutions embraced fair-pay guidelines, the work was typically underpaid.

Nonetheless, the platform attracted hundreds of thousands of workers, many of whom, like me, relied on AMT as a full-time job. I’ve spoken at length in many venues about 

. The short version is that I experienced a life-altering illness and needed to work from home. This was 2015, when remote work was far less common than it is today.

Over the years, Amazon and its supporters have argued that Mechanical Turk was never intended to provide a full-time job. Others have suggested that workers were simply choosing to accept low pay. I can say from experience that I had no meaningful choice.

My story is not unique. This work was marketed to single parents, caregivers, people with disabilities, and others who needed flexible work but faced barriers to traditional employment. Mechanical Turk offered that flexibility, but it was at a cost. Workers were expected to bear the platform's risks while having little power over its rules.

At the beginning of 2020, I became an organizer with 

, a worker-run, worker-led advocacy organization that helped workers navigate the multitude of problems that arose on AMT. These included account suspensions caused by faulty algorithms, mass rejections of completed work, and archaic security mechanisms that could ban workers who shared an IP address, including family members living in the same household. Organizers also met with AMT teams to deliver worker feedback and propose solutions to platform problems.

Later, as a community-based researcher with the 

, I focused on worker exploitation through the 

 project, helping to bring workers’ experiences and stories to public attention. One thing is very evident: the people doing the labor that made today’s automations possible are the least visible in the system.

Now, fast-forward to the Aug. 25 headlines about AMT shutting down. Only a fraction of workers and requesters were shown a bright yellow banner announcing the closure at the top of their screens. Instead, the news spread through online worker groups and forums. Most workers I encountered were finding out the news from one another 

“Mechanical Turk will close September 30, 2026.

We regularly evaluate our programs, tools, and services and make adjustments based on those assessments. Following an assessment, we've made the decision to close Amazon Mechanical Turk, effective September 30, 2026. Amazon Mechanical Turk will permanently close on September 30, 2026. For Workers and Requesters currently using the service, please visit our FAQ page to learn how you can prepare for this closure.”

Two days later, on Aug. 27, a bulk email arrived from Amazon. “Thank you for being part of Amazon Mechanical Turk's journey and for helping shape the crowdsourcing space,” it began, before announcing the closure and urging everyone to update their payment info. “We truly appreciate your contributions to MTurk over the years,” it ended.

In one month, workers scattered across more than 200 countries will lose access to income they relied on. Some have spent decades building their livelihoods around this platform. Hundreds of thousands of workers who have remained largely invisible while producing the data, evaluations, classifications, transcriptions, and other forms of labor that helped power today’s AI economy will be left without the platform—and meager income—it provided.

And this raises a much bigger question: Who is responsible for the people whose labor makes so-called technological progress possible?

Have you ever wondered why so many of the world’s largest technology companies are based in the United States, while so much of the labor supporting them takes place elsewhere? Companies including Amazon and Meta have built global systems that allow them to source digital labor across borders, often at extremely low rates. The language surrounding “data work” and “microtasks” has helped make this labor appear fundamentally different from other forms of employment, while the absence of comprehensive legislation has left workers with few meaningful protections.

There have been efforts to change this. In the US, Senator Ed Markey, D-Mass., and Representative Pramila Jayapal, D-Wash., have been among the lawmakers calling attention to the treatment of data workers and digital laborers. But congressional attention tends to focus on AI’s effects on traditional workplaces and employment. This work matters, but it leaves a glaring gap: Who is protecting the workers AI systems depend on in the first place?

The fact that there is still no comprehensive federal framework protecting data workers reveals a fundamental problem. The government does not adequately consider data work or the labor that underpins AI. Without even a consistent understanding of who these workers are, what they do, and how much their labor contributes to the economy, how can we expect policymakers to regulate the industry responsibly?

It is imperative that citizens contact their representatives and tell them that “AI” is about more than data centers, chatbots, and workplace automation. Behind every seemingly autonomous system are people whose labor is hidden from view.

The workers of Amazon Mechanical Turk have been part of the AI economy for more than two decades. They helped build the infrastructure that made today’s technology possible, often without recognition, security, or a meaningful voice in the systems they helped create.

And the question we should be asking is not simply what comes after Mechanical Turk. It is what kind of technological economy do we want to build next—and who gets to govern it?

Flock Safety camera on Haverhill Drive near Phillips Avenue in Toledo, Ohio. (Photo by Stephen Zenner / SOPA Images/Sipa USA)(Sipa via AP Images)

Unless Flock is Reined In, Mass Video Surveillance Will Grow Even More Pervasive

CEO of Meta Platforms Mark Zuckerberg attends the Allen & Company Sun Valley Conference at the Sun Valley Lodge on July 9, 2026 in Sun Valley, Idaho. (Photo by Kevin Dietsch/Getty Images)

Four States Take Meta to Trial in California, the Latest Front in Sprawling Multidistrict Fight

Cemetery for people on the move in Soufli, Evros, Greece. Photo by Petra Molnar.

The Racial Necrotechnics That Lead to Death By Design at Automated Borders

Tech Policy Press - Technology and Democracy

Tech Policy Press is tracking laws and regulations, along with government investigations and litigation, that will shape the rules and accountability for tech companies.

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe