Platforms Lose Consent

If You Only Read One Thing

The important word today is not automation. It is consent: Gig Work Gets a State shows Massachusetts moving platform labor from app terms to sector bargaining, while The Pope Becomes a Regulator shows AI governance borrowing moral authority from outside the labs. Start with the Vatican's Magnifica Humanitas, because it names the power problem plainly.

Gig Work Gets a State

Uber and Lyft spent a decade arguing that app labor is too fluid for old labor institutions. Massachusetts just answered with a new institution.

The state's Department of Labor Relations said on May 15 that the App Drivers Union had been designated by at least 25% of active rideshare drivers, triggering the final certification process under Chapter 150F. The same state explainer says the law lets a union become the exclusive representative either through an election or by showing 25% support, and that unionization does not make drivers employees. CBS Boston reported that the newly certified union will cover roughly 70,000 Massachusetts rideshare drivers, 18 months after voters approved the ballot measure.

Why it matters: This is not a standard "gig workers become employees" story. It is more interesting than that. Massachusetts is building a sector bargaining system around the fact that app drivers remain independent contractors, which means the fight moves from classification to rule-setting. The platforms keep flexible labor supply, but they lose the ability to make every pay, deactivation, and appeal question an individual contract term between an app and a driver. That is the structural break: consent is no longer a button tap in a terms-of-service flow; it becomes an administrated market process with a union, the Department of Labor Relations, and ultimately the state labor secretary in the loop.

The reason this travels beyond Massachusetts is that it offers a third path for states that do not want California's Prop 22 bargain and cannot easily force full employment classification. If the model works, the next policy question is not whether gig workers are employees in the old binary. It is which platform-dependent workforces can bargain by sector while preserving some contractor status. That threatens platforms less at the legal ontology layer and more at the operating layer: deactivation appeals, fare transparency, waiting time, data access, and algorithmic discipline.

Room for disagreement: A 25% designation threshold is a thin form of worker mandate, and companies may accept it precisely because it preserves contractor classification. A weak first contract could become a pressure valve rather than a power shift: drivers get a formal bargaining channel, while platforms keep the economic model intact.

What to watch: Watch the first six-month bargaining window. If the agreement covers deactivation appeals, fare-share transparency, and waiting-time economics, Massachusetts has created exportable platform labor infrastructure; if it stops at symbolic process language, the platforms bought classification peace cheaply.

The Pope Becomes a Regulator

The oddity was not that a pope wrote about technology. The oddity was that an Anthropic co-founder stood inside the Vatican while the pope warned that AI power is concentrating in private hands.

Pope Leo XIV presented Magnifica Humanitas on May 25, his first encyclical, focused on "safeguarding the human person" in the age of artificial intelligence. Vatican News had flagged the unusual lineup: cardinals and theologians alongside Christopher Olah, Anthropic co-founder and head of AI interpretability research. In the official presentation, Leo thanked Olah and said the church wanted to join the conversation while bringing "wisdom concerning the human" rather than technical answers.

Why it matters: AI regulation has a capacity problem. Governments can write procurement rules, safety institutes can test models, and courts can hear copyright and liability cases, but frontier labs still own most of the data, talent, compute, deployment telemetry, and failure evidence. That asymmetry gives companies enormous agenda-setting power before formal law catches up. The Vatican is not a regulator in the legal sense, but it is a legitimacy institution: it can convert abstract worries about concentration, labor, autonomy, weapons, and algorithmic exclusion into language that bishops, universities, unions, public agencies, and politicians can reuse.

That makes Olah's presence the structural signal. Anthropic benefits from appearing willing to submit to an outside moral voice, especially after months of AI policy stories turning on lab self-governance, procurement, and military use. The church benefits from showing that its social doctrine can speak to modern infrastructure, not just industrial-era factories. The phrase "soft law" usually means nonbinding norms that harden later through standards, procurement language, board policy, and public pressure. Magnifica Humanitas is a soft-law move aimed at the legitimacy layer of AI, and that is exactly where frontier labs are vulnerable.

Room for disagreement: The obvious criticism is that this is theater. Encyclicals do not audit models, regulate data centers, or force labs to disclose safety evidence. The worse version is moral laundering: companies stand beside institutions with spiritual authority while continuing to optimize for scale, retention, and compute advantage.

What to watch: Watch whether the encyclical's language on concentration, contestability, human dignity, and autonomous weapons appears in EU, state, university, or lab-governance documents by the end of September. If it does, the Vatican has moved from commentary to institutional vocabulary.

The Contrarian Take

Everyone says: The backlash against platforms and AI is finally arriving: workers are organizing, regulators are fining, the pope is warning, and X is policing creator theft.

Here's why that's wrong, or at least incomplete: This is not rejection. It is adoption reaching the point where institutions demand bargaining rights over systems they now depend on. Drivers still want rideshare work. Small businesses still buy ad tech. Courts still receive AI-assisted filings. Creators still need X distribution. The common move is that consent is being pulled out of private platform interfaces and pushed into state agencies, church doctrine, antitrust law, court administration, and creator attribution systems.

Under the Radar

The FTC case was about fake surveillance as much as privacy. The FTC said Cox Media Group, MindSift, and 1010 Digital Works will pay $930,000 over an "Active Listening" ad service that allegedly did not use voice data at all, but was sold as if it detected consumer conversations from smart devices. The important part is the consent theory: the agency said mandatory app terms are not real opt-in consent for such an invasive service even if the service had worked. (Source)
AI is lowering the cost of going to court and raising the cost of running courts. A new SSRN paper covering 4.5 million federal civil cases says non-prisoner pro se filings rose from an 11% long-run average to 16.8% in FY2025, first-180-day docket activity is up 158% from pre-AI means, and more than 18% of sampled 2026 complaints flagged as AI-generated. Access improved; court capacity did not. (Source)

Quick Takes

Google's DMA bill is becoming a product-design fight. Reuters, citing Handelsblatt, reported the EU is preparing a high triple-digit-million-euro fine over concerns that Google favors its own services in search results. Google's counter is revealing: it says DMA changes already made Search worse in Europe. That is the remedy fight in one sentence: competition law now directly rewrites interface quality. (Source)
Europe's cloud sovereignty package is really a procurement fight. The Atlantic Council notes the Commission's June 3 Tech Sovereignty Package is expected to include the Cloud and AI Development Act, a Chips Act update, and a definition of digital sovereignty. Definitions matter because they decide whether "sovereignty" means building European capacity or excluding non-European vendors from public-sector workloads. (Source)
X is discovering that creator payouts need property rules. X is cutting payouts for accounts that programmatically reupload smaller creators' videos and redirecting attribution toward originals. The lesson is not about one platform's moderation taste. Once distribution includes revenue share, reposting stops being a norm problem and becomes market plumbing. (Source)

The Thread

Today's thread is consent moving up the stack. Massachusetts moved driver consent from individual app contracts to state-supervised bargaining. Pope Leo moved AI consent from lab self-assurance to moral doctrine. The FTC treated buried app terms as insufficient for intimate data claims. Courts are absorbing the costs of AI-assisted access. X is converting creator attribution into payout logic. The pattern is simple: when a platform becomes infrastructure, private consent screens are no longer enough to legitimate the market.

Predictions

New predictions:

I predict: By 2026-08-31, at least one state outside Massachusetts will formally advance a gig-worker bargaining bill or rule that preserves contractor status while creating a sectoral bargaining process modeled on Massachusetts. (Confidence: medium; Check by: 2026-08-31)
I predict: By 2026-09-30, at least one major AI lab or public-sector AI procurement body will cite Magnifica Humanitas or a Vatican-led AI dialogue in a governance, safety, or policy document. (Confidence: medium; Check by: 2026-09-30)

2026-05-26 03:20 EDT