Industries / Government, education and community / Government / Courts / Arbitration
Consumer and employment cases
Arbitration clauses that businesses write into standard contracts for customers and workers, and the courts, lawmakers and providers that decide how far those clauses reach.
7
Milestones achieved, from 1991 to 2026.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1991
Job bias claims arbitrated
In Gilmer v. Interstate/Johnson Lane, decided on 13 May 1991, the US Supreme Court held that a claim under the federal age discrimination law could be sent to binding arbitration under an agreement in a stockbroker's registration form. The Court said that unequal bargaining power between employers and employees is not by itself a reason to refuse to enforce such agreements.
Credit
US Supreme Court, opinion by Justice Byron White
2011
Class waivers upheld
A class waiver is a clause that bars a customer from joining a group lawsuit or a group arbitration. In AT&T Mobility v. Concepcion, decided on 27 April 2011, the US Supreme Court held that the Federal Arbitration Act overrides a California rule that treated such waivers in consumer contracts as unconscionable, and in Epic Systems v. Lewis, decided on 21 May 2018, it held that waivers in employment arbitration agreements must also be enforced.
LessonA rule made by an agency can be undone within months: the Consumer Financial Protection Bureau published a rule on 19 July 2017 that barred such waivers in many consumer finance contracts, and Congress voided it in a resolution approved on 1 November 2017.
Credit
US Supreme Court
2013
Europe keeps courts open
The EU directive on consumer alternative dispute resolution (ADR), dated 21 May 2013, required member states by 9 July 2015 to ensure that any consumer dispute with a trader based there can go to an ADR body that meets common quality rules, and that such bodies normally give an outcome within 90 calendar days. It also made an agreement to use ADR not binding on the consumer if it was made before the dispute arose and would stop the consumer from going to court.
Credit
European Parliament and Council of the EU
2020
Mass arbitration arrives
Mass arbitration means filing thousands of separate claims at once, each of which triggers fees the business agreed to pay. In 2019 lawyers for DoorDash couriers filed about 6,250 individual claims with the American Arbitration Association (AAA); DoorDash declined to pay "nearly $12 million" in fees, and in February 2020 US District Judge William Alsup ordered it to arbitrate the claims of 5,010 couriers, writing that the company "now blanches at the cost of the filing fees it agreed to pay".
LessonCounter-moves written by businesses can fail as well: on 28 October 2024 the Ninth Circuit refused to enforce Ticketmaster's clause sending claims to New Era ADR, whose rules applied rulings in test cases to later claimants, gave no right to discovery and limited appeals.
Credit
Judge William Alsup, US District Court for the Northern District of California; the law firm Keller Lenkner, for the couriers
2020
Canada strikes a fee wall
Uber's standard contract required drivers to resolve disputes through mediation and arbitration in the Netherlands, with up-front fees of US$14,500. On 26 June 2020 the Supreme Court of Canada held the clause unconscionable for a Toronto food delivery driver who earned $400 to $600 a week, because the fees were most of his annual income.
Credit
Supreme Court of Canada; David Heller, the driver who brought the case
2022
Harassment claims leave arbitration
The Ending Forced Arbitration of Sexual Assault and Sexual Harassment Act, which became US law on 3 March 2022, lets a person bringing a sexual harassment or sexual assault case choose court even after signing an arbitration agreement or a waiver of group actions before the dispute arose.
Credit
US Congress
2026
Local delivery drivers exempted
The Federal Arbitration Act of 1925 does not cover the employment contracts of "seamen, railroad employees, or any other class of workers engaged in foreign or interstate commerce." On 28 May 2026, in Flowers Foods v. Brock, a unanimous US Supreme Court held that this exemption is not limited to workers who cross state lines, so a driver who carries goods on the in-state leg of an interstate journey can qualify.
Credit
US Supreme Court, opinion by Justice Neil Gorsuch
Watch for
A person reviews automated outcomes
A directive amending the EU's consumer ADR rules, adopted on 16 December 2025, says parties must be told in advance when an ADR body uses automated means to reach an outcome, and may ask for that outcome to be reviewed by a person. It also lets traders outside the EU take part, and gives EU traders 20 working days, at most 30, to say whether they will join a case.
Achieved whenmember states apply the new rules, which are due from 20 September 2028, and ADR bodies' reports show how many automated outcomes were reviewed.
Claimed by
European Parliament and Council of the EU, December 2025
Mass cases settle early
The AAA reports that 72% of the 6,360 consumer mass-arbitration cases closed at the merits stage in 2025 settled, that 96% of settled consumer cases did so before a merits arbitrator was appointed, and that individual demands filed in mass arbitrations fell from about 280,000 in 2024 to just under 105,000 in 2025. It suggests that early rulings by its "process arbitrator", one arbitrator who decides filing and other non-merits questions for all the claims, may be helping parties settle.
Achieved whena study by someone outside the AAA, using records from more than one provider, measures whether the process-arbitrator stage shortens mass cases.
Claimed by
American Arbitration Association, June 2026
Open
01
Sign-up screens measured
A public dataset that records how the 200 most-used consumer apps and websites in the United States show their terms, which usually contain an arbitration clause, at sign-up or checkout: whether a box must be ticked, the distance in pixels between the notice and the button, the notice's font size against the button text, the contrast ratio of the link, and how many screens the flow has. It measures design only and says nothing about whether any contract binds anyone.
The briefCredit · unclaimed02
Europe's consumer ADR scorecard
One open table of the activity reports that EU consumer ADR bodies must publish, covering the number of disputes received, the share refused and why, the share discontinued, the average time to resolve and the rate of compliance with outcomes, standardized across languages and linked to each source page, with a count of how many bodies report an average within the 90-day limit.
The briefCredit · unclaimed
Open 01 · the brief
Sign-up screens measured
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public dataset that records how the 200 most-used consumer apps and websites in the United States show their terms, which usually contain an arbitration clause, at sign-up or checkout: whether a box must be ticked, the distance in pixels between the notice and the button, the notice's font size against the button text, the contrast ratio of the link, and how many screens the flow has. It measures design only and says nothing about whether any contract binds anyone.
Why it is still open
Courts decide these cases one interface at a time on the same features, such as font size, color, placement next to the button and competing elements, and reach different results, as in Tejon v. Zeus Networks (Eleventh Circuit, 2026) and Dahdah v. Rocket Mortgage (Sixth Circuit, 2026). Services such as ConductAtlas track the text of terms across hundreds of platforms, but a search for this research found no published measurement of how sign-up screens present them.
Built with what exists
Browser Use 1.0 or a Chrome DevTools Protocol script to walk each flow and save screenshots and page structure (the Note "Deleting 80% to start over" covers Browser Use); axe-core, Deque's open-source accessibility checker, for contrast ratios under WCAG 2.2; Python for the measurements. Test e-mail addresses only, and every flow stops before any payment.
First step
Record the sign-up flows of 20 popular apps this month, measure the five features by hand, and fix the written protocol before automating it.
Done looks like
200 flows recorded with screenshots and the five measurements; 20 flows measured by hand by two people, with their agreement with the software reported; the run repeated three months later with every change listed.
Open 02 · the brief
Europe's consumer ADR scorecard
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
One open table of the activity reports that EU consumer ADR bodies must publish, covering the number of disputes received, the share refused and why, the share discontinued, the average time to resolve and the rate of compliance with outcomes, standardized across languages and linked to each source page, with a count of how many bodies report an average within the 90-day limit.
Why it is still open
The European Commission's 2023 proposal to revise the directive drew its figures from 23 national authorities, about 300,000 eligible disputes a year across the EU, and said that a lack of cost data makes the cost-effectiveness of ADR hard to calculate; a search for this research found no table that compares bodies one by one. From 2028 the reports will be due every two years instead of every year, so a baseline is easier to build now.
Built with what exists
Each member state's published list of ADR bodies; Python to collect reports; Poppler's pdftotext and Tesseract OCR for scanned files; a translation model; a large language model to pull the eight required figures, every figure checked against its page.
First step
Take one member state's list of ADR bodies, collect their latest reports and tabulate the eight figures by hand.
Done looks like
Reports from at least 300 bodies in all 27 member states in one public file; 10% of figures checked by hand and the error rate published; a short summary of how many bodies publish each required figure at all.