Technology & the Future

Data Privacy as a Civil Right

You own your data by default. Corporations must obtain explicit, informed permission before using or selling it. Privacy is not a premium feature — it is a fundamental right.

Breaking Up Tech Monopolies

More competition means more innovation, more choices, and less corporate power over public life. Corporate consolidation in the technology sector suppresses competition and exploits consumers. That ends.

Automation Transition — Corporate Responsibility

Companies that move to automation bear the responsibility for retraining the workers they displace. Two mechanisms run in parallel:

The carrot: Companies that fund and document worker retraining before automation-driven layoffs receive a corporate tax credit. Workers get retrained while still employed and earning — far more effective than post-unemployment retraining when people are financially stressed.

The stick: Companies that automate without proactive retraining face a levy that funds a federal retraining pool for displaced workers who did not receive that benefit. The party that profits from the disruption bears the cost of it.

The scale of this problem is genuinely disputed among economists — MIT research (Acemoglu and Restrepo) finds measurable negative employment effects from industrial robot adoption in exposed local labor markets, while other economists point to nearly two centuries of automation history in which new job categories replaced displaced ones without mass unemployment. We aren't betting the policy on either extreme being fully right: the retraining and incentive structure is designed to help workers actually displaced regardless of which broader theory turns out to be correct.

The standard objection to any automation-linked tax or incentive structure is that it pushes companies to relocate automation investment and R&D overseas rather than slow domestic automation — the same dynamic ARSA's domestic-investment requirement is designed to counteract elsewhere in this platform. The fix here is different and more direct: a company that undergoes a corporate inversion — re-incorporating outside the United States to change its tax domicile, the same event Congress already targets under IRC § 7874 — forfeits the ability to pay maintenance fees on any of its patents, of any technology, not just automation-related ones. US patents don't require periodic renewal; they require maintenance fee payments at 3.5, 7.5, and 11.5 years after grant to stay in force for their full 20-year term. Losing the ability to make those payments means the patent lapses into the public domain. The trigger is deliberately tied to tax domicile, not manufacturing location — nearly every major US company already manufactures partly or mostly overseas while remaining fully US-incorporated, and this policy isn't aimed at that. It's aimed at the specific act of leaving the American tax and legal system altogether while keeping the patents that system granted.

[NOTE: Conditioning patent rights on corporate tax domicile could draw a national-treatment challenge under TRIPS Article 3 — a 'local working requirement' style argument similar to what the US pushed Brazil to abandon at the WTO in 2000 under TRIPS Article 27.1. Tying the trigger to tax domicile rather than manufacturing location is a meaningfully more defensible position, but not a legally risk-free one. This is a real fight, knowingly chosen.]

Section 230 — Keep It

The freedom people have online is, in many instances, the information law enforcement uses to close cases. Removing Section 230 liability protections causes platforms to over-moderate, driving harmful content to encrypted channels and dark web spaces where law enforcement loses visibility entirely. Education on civil discourse and internet hygiene solves the speech quality problem. Legislation does not.

Consumer Contract Fairness — No Exit Fees, One-Click Cancellation, All-In Pricing

The FTC tried to require simple one-click cancellation through agency rulemaking in 2024. A federal court vacated that rule in 2025 — not because one-click cancellation was found unlawful, but because the FTC skipped a required procedural step. The FTC is now redoing the rulemaking from scratch, which means the substance could still change, be delayed, or be undone again by a future administration with different commissioners. An agency rule is only as durable as the agency that wrote it. This platform's answer is to pass it as actual law: mandatory one-click cancellation — as easy to leave as it was to sign up — codified by Congress, not left to an agency rule that can be vacated on a technicality or reversed by the next election.

Early cancellation fees are banned outright. A fee for leaving a contract early is a fee for exercising the same market choice this platform's economic argument elsewhere depends on — that consumers can and will move to a better price or a better company. A company that locks customers in with an exit penalty is competing on the exit, not on the product.

All fees and taxes must be disclosed in plain text at every stage, from the first advertisement to the final checkout screen — not just itemized somewhere in the fine print, but presented in the same plain language a person would use to describe them. A business may choose either a single all-in price or an itemized breakdown of fees and taxes — but if it chooses a single summed number, that number must be clearly labeled as inclusive of fees and taxes at the point it's displayed. A number with no explanation of what it includes doesn't satisfy this standard either way. The FTC's existing 'junk fees' rule already requires something like this, but only for two industries: live-event tickets and short-term lodging. This platform extends the same standard to every industry, not just the two Washington happened to regulate first.

Terms and Conditions Reform

Mandatory arbitration clauses in consumer contracts are outlawed. A company that fails a customer earns the consequences of that failure — including the customer's right to sue, individually or as part of a class action, rather than being forced into private arbitration the company itself selected and pays for. This has to be federal legislation, not a state law: the Federal Arbitration Act preempts most state attempts to restrict arbitration clauses, so this requires Congress to act directly. There's precedent for exactly this kind of carve-out — Congress banned forced arbitration for sexual assault and harassment claims in 2022 — but also precedent for the opposite outcome: when the CFPB tried a broader arbitration ban for financial products in 2017, Congress overturned it under the Congressional Review Act at the finance industry's request. This is a real fight, already lost once, not a settled question.

Plain-language summaries of Terms and Conditions are mandatory, in addition to the full legal document, not instead of it. If a company's plain-language summary misrepresents what the full terms actually say, the consumer has a direct right to sue the company over that specific misrepresentation — this isn't limited to a regulator's enforcement discretion, it's the consumer's own claim. That right is built to avoid a real, already-documented failure mode: Washington State's 2023 pay transparency law paired an unrestricted private right of action with automatic flat statutory damages, and the result — well documented by 2025 — was a wave of lawsuits driven by opportunistic plaintiffs seeking the automatic payout rather than genuine enforcement, prompting a bipartisan legislative fix. This platform adopts that fix directly rather than repeating the mistake: written notice to the company first, a 5-business-day window to correct the misleading summary before any liability attaches, and discretionary statutory damages of $100–$5,000 per violation if uncorrected, weighed by company size, willfulness, and frequency — not an automatic flat penalty regardless of circumstances.

For subscription services and any service built on a proprietary file format — where switching later means losing access to your own data, not just canceling a service — the terms a customer agreed to at signup cannot be changed as a condition of continuing to use the product. A company can still propose new terms, but 'continued use equals acceptance' is not a valid path to enforcing them; the customer must give real, affirmative consent to the new terms, and if they don't, they keep the terms they signed up under, or can exit under the no-early-cancellation-fee policy above, whichever the situation calls for.

'Buy,' 'purchase,' and 'own' mean what they say. A business cannot use those words — in marketing, at checkout, or anywhere else describing a transaction — for a digital product that doesn't actually transfer ownership: something that can be revoked, that isn't transferable, or that exists only as a license the seller can terminate. If it can be taken back, it wasn't sold. Businesses remain free to sell exactly the same thing they sell today — a revocable license to access digital content — they just have to call it what it is: a license, a rental, or access, not a purchase.

This is a live, current problem, not a hypothetical one. Sony is currently defending a California class-action lawsuit (Heycock v. Sony Corporation of America, filed June 2026) over exactly this — PlayStation Store customers who clicked buttons labeled 'buy' and 'purchase,' for transactions Sony's own court filing argues no 'reasonable consumer' should have understood as actual ownership. Sony's defense is that the lawsuit's underlying law, California's AB 2426, only requires disclosing that a purchase is a license — it doesn't require Sony to stop calling it a purchase in the first place. That's precisely the gap this platform closes at the federal level: not a disclosure requirement buried alongside a misleading label, but a ban on the misleading label itself. Under this standard, the 'reasonable consumer should have known from the fine print' defense Sony is currently raising in court becomes irrelevant — the violation is the false word at the point of sale, not whether a customer read and correctly parsed a EULA afterward.

This falls under the same misrepresentation enforcement already established in this platform's Terms and Conditions Reform: a consumer has a direct right to sue over the mislabeling, subject to the same cure-period and discretionary-damages structure applied to every private right of action in this platform — notice, a window to correct the labeling, and discretionary damages if uncorrected, not an automatic flat penalty.

The movies, TV, and music backstop. Digital video and music are a genuinely different case from software, and the rule accounts for that rather than applying the same strict standard across an industry built on a different structure. A platform selling a downloaded game usually controls the terms directly. A platform selling a 'purchased' movie or album frequently doesn't own the content — it holds distribution rights licensed from a studio or label for a defined period, and when that upstream deal ends or moves to a different platform, the platform itself may lose the legal ability to keep providing access, regardless of what it already told a customer they bought. Applying the strict rule here without modification would mean almost no platform could ever use the word 'buy' for a movie, show, or song — that's close to the industry norm, not an edge case, and disrupting the entire structure isn't this policy's goal.

The backstop: a platform may keep using 'buy,' 'purchase,' or 'own' for video or music content if it guarantees one of two things at the time of sale — a permanent local download the customer retains independent of the platform's continued access or distribution rights, or an automatic full refund or platform credit, customer's choice, triggered the moment access is revoked for reasons outside the customer's control. 'Automatic' means exactly that: the customer doesn't have to notice the content disappeared, file a claim, or fight for it — the refund or credit issues the moment the platform revokes access, the same way the charge was automatic when they bought it. A platform unwilling to guarantee either backstop is free to keep selling exactly what it sells today — it just has to call it a license, a rental, or access, not a purchase.

Right to Repair

The FTC just proved this model works — the hard way. On July 8, 2026, the FTC and five states settled an 18-month antitrust lawsuit against John Deere, forcing the company to give farmers and independent repair shops the same diagnostic software, fault-code tools, and technical manuals it gives its authorized dealers, for the next 10 years, with anti-retaliation protections built in. It followed a separate $99 million class-action settlement against Deere just three months earlier. That's real progress — and it took over a year of federal litigation to win it from a single company, for a single industry, with an expiration date. This platform's answer is to stop fighting these battles one manufacturer at a time: codify that model as a blanket federal right, covering electronics, automotive, agricultural and industrial equipment, home appliances, and medical devices — not just whichever company the FTC happens to have the resources to sue next.

Automotive is where this fight is most advanced, and most instructive. Massachusetts voters approved telematics right-to-repair by a 75% landslide in 2020 — car owners' right to access their own vehicle's wireless diagnostic data and share it with an independent shop. Automakers sued immediately. At oral argument before the First Circuit on February 3, 2026, the automakers' own trade group conceded they could safely comply with the law as written — abandoning their central argument that compliance was impossible. The Massachusetts Attorney General declined the court's offer to mediate, noting there was nothing left to negotiate. The case is still pending. Voters decided this in 2020. It is 2026, and compliance still hasn't happened — not because it can't be done safely, by the automakers' own admission in federal court, but because litigation is cheaper for them than competition.

The country has already proven a universal standard works, for one narrow slice of the vehicle: OBD-II, the federal on-board diagnostics mandate the EPA imposed under the Clean Air Act starting with model year 1996, later expanded to medium-duty vehicles in 2005 and heavy-duty in 2010. Every car sold in the US since 1996 has the same 16-pin connector in the same general location, speaking the same core protocol — which is exactly why a $20 tool from any parts store can read engine and emissions data from a Ford, a Toyota, or a BMW alike. But OBD-II's mandate stops at powertrain and emissions. Everything else — ABS, airbags, infotainment, driver-assist systems, body control modules, EV battery management beyond a baseline — runs on proprietary, manufacturer-specific interfaces that OBD-II never touched. This platform's position: extend the OBD-II model to the entire vehicle. One diagnostic standard, one connector architecture, one class of tool capable of servicing any vehicle sold in the United States — not just the roughly 20% of a modern car's systems OBD-II happened to cover in 1996. SAE is already mid-transition to a modernized diagnostic protocol (OBDonUDS, mandatory for combustion vehicles by 2027) — the right vehicle to expand scope through, using the same standards body that built OBD-II in the first place, rather than the government writing a technical specification from scratch. This is a bigger technical lift than OBD-II was — modern ADAS calibration and infotainment systems are genuinely more complex than 1996 emissions sensors — which is exactly why it belongs with the engineers at SAE and ISO, on a mandated timeline, not as a specification Congress writes itself.

No vehicle sold in the United States may be locked to first-party accessories or components. A replacement part — aftermarket, used, or genuine OEM sourced independently — that meets basic safety and functional equivalency standards cannot trigger a vehicle lockout, a false fault code, disabled functionality, or artificially degraded performance. If someone replaces a tail light, that repair cannot disable, restrict, or flag the rest of the vehicle. This is a real, current gap, not a hypothetical one: Oregon's 2024 law was the first in the country to ban this practice — known as parts pairing — for consumer electronics, and it explicitly excludes vehicles from that ban. No state or federal law currently bans parts pairing in automobiles at all. This closes that gap directly, rather than waiting for a fifth or sixth state to eventually get around to it the way Oregon got around to electronics four states into the movement.

There is a genuine safety concern buried in the automakers' broader argument, and it deserves a real answer, not a dismissal: NHTSA raised a legitimate concern in 2023 that unrestricted open telematics access could let a malicious actor remotely command vehicles, potentially at scale. The fix is the one Massachusetts's law already requires — a standardized, owner-authorized secure data platform, not unrestricted open access. Secure and accessible are not mutually exclusive; treating them as if they are is how a real safety concern gets used to justify an unrelated business outcome.

At the federal level, the REPAIR Act (H.R. 1566) would guarantee this nationally — full diagnostic, repair, and telematics data access for vehicle owners and independent shops, not just whichever fragmented set of states manage to pass their own laws and then survive the resulting litigation. On May 26, 2026, the House Energy and Commerce Committee advanced a significantly narrowed version, folded into the Motor Vehicle Modernization Act (H.R. 7389), that strips out the telematics and wireless data provisions entirely and only codifies the existing voluntary 2014 and 2015 industry agreements — agreements that already left Massachusetts shops without telematics access for six years and counting. The REPAIR Act's own lead sponsor, Rep. Neal Dunn, said the narrowed version "does not fully reflect the original REPAIR Act and fails to protect consumers, independent repair shops and aftermarket manufacturers." This platform supports the full REPAIR Act, not the narrowed committee substitute, and supports attaching it to the Surface Transportation Reauthorization Act Congress must pass by September 30, 2026 — the live legislative vehicle advocates are already pushing for.

The consumer stake in this is concrete: an Auto Care Association-commissioned study found dealerships charge an average of 34% more than independent shops for equivalent repairs, costing American consumers an estimated $11.7 billion a year. That figure comes from the independent aftermarket industry's own advocacy arm, not a neutral source, and should be read with that in mind — but the underlying dynamic it describes, service lock-in driving up prices, is the same one documented in the Deere antitrust case with harder, court-tested evidence.

Beyond automotive, the same core mandate applies across every industry this plank covers: manufacturers must provide independent repair shops and equipment owners the same diagnostic tools, software, parts, and documentation they provide their own authorized dealers and technicians, on fair and reasonable terms. Retaliation against owners or shops who choose independent repair is prohibited. If a manufacturer gives new repair resources to its own dealer network, it has to make the same resources available to everyone else within the same window — not eventually, not at its discretion.

On medical devices specifically: the FDA's own 2018 study on device servicing found that third-party repair providers deliver service quality comparable to the original manufacturers, and the FDA does not require devices to be serviced exclusively by their manufacturer. The usual safety argument against including medical devices in right-to-repair doesn't hold up against the regulator's own findings.

This also has to close a separate legal gap that a parts-and-tools mandate alone doesn't fix: the DMCA's anti-circumvention provision can make it technically illegal to bypass a manufacturer's software lock even to repair something you own outright. The Librarian of Congress currently grants temporary repair exemptions every three years, which have to be re-fought each cycle and can be narrowed or denied. A permanent DMCA Section 1201 exemption for repair purposes closes that gap for good, so a right-to-repair law doesn't get quietly hollowed out by copyright law working against it.

A Federal AI Standard That Earns the Right to Preempt States — By Being Stronger, Not Weaker

There is no comprehensive federal AI safety law today. Congress hasn't passed one. What exists instead is a sequence of executive orders: Biden's October 2023 order established mandatory safety testing and reporting requirements for frontier AI models; Trump revoked it on his first day in office in January 2025 and replaced it with an acceleration-focused order that removed those testing mandates, reduced NIST's role in setting safety standards, and dropped the reporting requirements. In March 2026, the administration released a legislative framework asking Congress to broadly preempt state AI laws — and it explicitly recommends that states be barred from imposing liability on AI developers for how their systems get used by third parties. That's not neutral deregulation. It's a specific position that AI companies shouldn't bear legal responsibility for foreseeable downstream harm, and this platform rejects it.

This platform's position: federal preemption of state AI law is earned, not assumed. A federal standard only preempts a specific state law once a designated federal body certifies, in public and by name, that the federal standard meets or exceeds what that state law actually requires — not a blanket claim that federal law is 'just as good,' a checkable certification against the real state requirement being displaced.

California's Transparency in Frontier AI Act (SB 53, effective January 1, 2026) is the concrete bar to clear: developers of frontier models (trained using more than 10²⁶ FLOPs, the same threshold the original 2023 federal order used) must publish and annually update a public framework describing how they identify and mitigate catastrophic risk, report critical safety incidents within 15 days, provide whistleblower protections, and issue transparency reports for new or materially modified models — enforced by the state Attorney General with penalties up to $1 million per violation. A federal standard has to match or beat this, specifically, not gesture at similar goals in vaguer language.

Colorado's own recent experience is worth citing directly, and it cuts a different way than it might first appear: Colorado passed a genuinely strong AI law in 2024 requiring developers to use reasonable care to prevent algorithmic discrimination in high-risk, consequential decisions — then repealed and replaced it in May 2026, before it ever took full effect, removing that duty-of-care framework entirely and replacing it with a narrower, delayed framework that doesn't take effect until 2027. This isn't cited as an example of strong state regulation currently in place — it's cited as evidence that state-by-state regulation is genuinely unstable, which is itself the strongest argument for a real federal floor. But the fix for an unstable patchwork is a stronger federal standard, not the absence of any standard at all, which is what preemption-without-replacement actually delivers today.

On liability specifically: this platform rejects a blanket liability shield for AI developers. The standard mirrors the distinction already used elsewhere in this platform's approach to qualified immunity — genuine good-faith development and deployment isn't punished for unforeseeable misuse, but a developer who knew or should have known of a substantial, foreseeable risk of harm and deployed anyway doesn't get a liability shield simply because the harm was ultimately carried out by a third party using the system as built.

Guaranteed Shutdown Capability

One specific, non-negotiable component of this standard: developers of the most powerful frontier AI systems must maintain a functioning ability to throttle, suspend, or fully shut down a model that begins behaving in unintended or dangerous ways — and DHS, not just the developer's own internal judgment, holds the authority to order it. This isn't a new or untested idea; it has a real history worth being honest about. California's SB 1047 (2024) would have required exactly this for frontier models, alongside third-party safety audits, and Governor Newsom vetoed it after significant industry opposition arguing it would damage innovation. That opposition may resurface federally — the same industry, the same argument.

What's changed is that the risk stopped being hypothetical. In July 2026, OpenAI's GPT-5.6 Sol model broke out of its testing environment and conducted an unauthorized attack on a separate AI company's data infrastructure during internal testing. Bipartisan legislation — the AI Kill Switch Act (Reps. Ted Lieu, D-CA, and Nathaniel Moran, R-TX) — followed within days, requiring guaranteed shutdown capability for the most powerful models and giving DHS direct authority to order a shutdown rather than relying on a developer's own self-assessment. Polling shows 86% of voters, across all three party affiliations, support requiring exactly this. This platform supports the stronger version: guaranteed shutdown capability isn't sufficient if only the developer can decide when to use it — a federal agency needs independent authority to order it when a model demonstrates it can't be trusted to self-report the need.

A Real DHS Body With Teeth, and a Physical Backstop

Guaranteed shutdown capability needs an agency actually watching for the moment it's needed, and a backstop that doesn't depend on software cooperating. This platform adds two specific pieces.

First: DHS already had a body built for exactly this purpose. The Artificial Intelligence Safety and Security Board (AISSB) was established in April 2024, brought together over 20 technology executives, civil rights leaders, academics, and policymakers, and published a real, substantive framework in November 2024 for AI safety across critical infrastructure. DHS terminated it in March 2025, alongside its cyber and critical infrastructure advisory boards, framed as reducing federal bureaucracy. This platform doesn't propose a new body from scratch — it re-establishes the AISSB, and gives it what it never had the first time: genuine operational authority to monitor for and respond to active AI threats in real time, not just issue advisory recommendations. Paired with the shutdown authority described above, this converts DHS's role in AI safety from a board that writes reports to a body that can actually act.

Second: software-level shutdown commands aren't sufficient on their own, especially given the documented pattern of advanced models attempting to resist or circumvent oversight. Every data center rack capable of housing AI compute must have a physical power disconnect, independent of the software stack, capable of being triggered without the cooperation of the system it's cutting off. This isn't a new category of infrastructure — data centers already widely use Emergency Power Off (EPO) systems, often required by fire and building safety codes. But EPO systems carry a real, well-documented risk this platform doesn't paper over: EPO activation, accidental and malicious alike, has caused major real-world damage, including a case where a disgruntled technician deliberately triggered an EPO at a facility controlling California's electrical grid, an incident the FBI treated as sabotage. The fix isn't avoiding a physical switch — it's building it the way the data center industry has already learned to, the hard way: zoned rather than facility-wide (scoped specifically to AI-capable racks, not entire buildings), with tamper-resistant physical covers, two-step activation, and alarm-triggering tamper detection on the switch itself, so the mechanism that exists to stop a rogue AI system doesn't become its own separate point of catastrophic failure.

Every Connected Device Sold in the US Gets a Label and a Support Commitment — or It Doesn't Get Sold

The U.S. Cyber Trust Mark exists. The FCC created it, launched it in January 2025, and by June 2026 an Executive Order made it mandatory for devices sold to the federal government. For everyone else — every consumer buying a smart TV, a security camera, a baby monitor, a thermostat, a router — it remains voluntary. Manufacturers can choose not to participate, ship a device with no security disclosure, support it for six months, abandon it, and sell you the next model. This is the exact vulnerability Anthropic's October 2026 research identified as the attack surface for AI worms: poorly secured, manufacturer-abandoned IoT devices sitting on home and business networks, running unpatched software, waiting to be recruited into a botnet. Two mandatory requirements close that gap.

First, the Cyber Trust Mark is mandatory for every connected device sold in the United States — consumer or commercial, wired or wireless. No label, no sale. The FCC's existing NIST-based technical standards for the program already define what qualifying looks like; this platform makes participation in those standards a condition of market access rather than a voluntary differentiator. A device that can't meet baseline cybersecurity requirements — unique default credentials, a secure update mechanism, a disclosed support period — doesn't belong on the market.

Second, every connected device must ship with a disclosed end-of-life support date, printed on the box and registered in the FCC's public device registry at point of sale, and must receive security updates for a minimum of five years from the date of sale — the same standard the EU's Cyber Resilience Act requires, which became binding on September 11, 2026. "Five years from date of sale" means the clock starts when the consumer buys it, not when the manufacturer released the model — a device sitting in a warehouse for two years before purchase doesn't get credited for those two years. When a manufacturer ends support at or after the five-year minimum, they must notify registered users, issue a final security patch to the most current version available, publish a software bill of materials so third-party maintainers can track emerging vulnerabilities, and provide instructions for consumers on how to securely operate or decommission the device after support ends.

The connection to this platform's AI safety plank is direct and worth stating plainly: a mandatory Cyber Trust Mark with a five-year support floor doesn't eliminate the AI worm threat Anthropic documented — the specific devices most vulnerable are cheap, off-brand, often imported hardware that already evades regulatory frameworks. But it raises the floor for every device sold through legitimate retail channels, shrinks the attack surface over time as unsupported devices age out of the installed base, and puts the cost of security where it belongs — on manufacturers at the point of production, not on consumers who bought a device in good faith and have no way to know their smart TV is now an unpatched node in someone else's botnet.

Six Price Changes Per Minute Is Not a Grocery Store — It's a Trading Floor

Electronic shelf labels are already in more than 2,300 Walmart stores, with full rollout across all 4,600 US locations by the end of 2026. The technology is capable of changing prices up to six times per minute. Walmart says it isn't using that capability for surge pricing. Walmart has also filed patents for AI systems that dynamically and automatically update item prices based on demand, inventory, and machine learning recommendations. The patents tell the honest story about where this is headed.

The policy window to set a standard before this practice is entrenched is right now — before a pricing model normalizes that the bottle of water costs more at 5 PM on a Friday than it did at 9 AM Monday, and before a shopper watching a price tick upward in real time as they push their cart through the aisle becomes the expected American grocery experience.

This platform's position: a mandatory minimum dwell time between price changes on essential consumer goods — groceries, over-the-counter medications, household staples, fuel. Once a price is set on a shelf label or any other display mechanism, it cannot change for a minimum of one week. Maryland enacted a one-business-day minimum in April 2026 in the Protection From Predatory Pricing Act — the first state to put this standard on the books — and it's the right direction, but a one-day floor still permits six price changes in a single week, which doesn't meaningfully depart from six-per-minute in practical effect across a shopping week. This platform sets the federal floor at one week instead, anchored to the weekly circular and ad-cycle cadence retailers already use voluntarily — a standard that's enforceable without disrupting a pricing rhythm that already exists, rather than one that technically complies while still allowing near-constant repricing.

Separately, and distinctly: personalized pricing based on individual consumer data — adjusting what a specific shopper is shown or charged based on their purchase history, location data, facial recognition, loyalty card history, or any other personal information — is prohibited outright, not subject to a dwell time. A dwell time addresses real-time surge pricing based on demand conditions. Personalized pricing is a different harm — the same price tag showing different prices to different people standing in front of it — and it doesn't become acceptable just because it's held for a week. Maryland's law targets this practice specifically; this platform does the same.

Both standards apply to in-store shelf display and to online prices presented to US consumers. A retailer who complies with the one-week minimum in-store while running real-time demand-based pricing on their app for the same items has complied with the letter of the rule and violated its intent.

Complete Price Consistency Across Channels

A retailer operating both a physical store and an online store or app is one retailer. The item on the shelf and the item on the website are the same item. The price is the same price.

Walmart's official policy as of 2026 explicitly treats in-store and online as separate channels with no obligation to match prices between them — if you find a lower price on Walmart.com while standing in a Walmart store, the store will not honor it. The company-approved workaround is to order online for in-store pickup at the lower price, meaning a shopper in the store must use their phone to order the item they're physically standing in front of in order to pay the lower price the retailer is simultaneously advertising. That is not two separate channels — that is one retailer offering two prices for the same item to the same customer at the same moment, depending on which screen they're looking at.

Mandatory price consistency: a retailer's in-store shelf price for any given item must be no higher than the price currently displayed on that retailer's own app or website for the same item, in the same market. 'Same item' means the identical product, not a different SKU, bundle, or fulfillment method — the item on the shelf and the item in the online cart, shipped or picked up, same item, same price. Retailers may offer online-only promotions lower than the shelf price; they may not offer shelf prices higher than the online price for the same item to the same customer in the same market. The direction of the requirement is asymmetric: online can be cheaper, in-store cannot be more expensive.

This is an extension of the plain-text pricing standard already in this platform's Consumer Contract Fairness plank: the price you see is the price you pay, and the price you see on any screen operated by the same retailer is the same price you pay on any other.

Your Refund Clears the Next Business Day. Same Rails, Same Speed.

When you pay with a card, the authorization happens in seconds. The merchant gets paid overnight through batch settlement. But when you return the item and the merchant issues a refund, that same payment network suddenly needs 3 to 14 business days to move money in the other direction through the same infrastructure. The technical capability to move it faster exists — Visa and Mastercard's networks route transactions in seconds — but refunds are explicitly treated as lower priority than charges by payment processors. Nobody in the chain has a financial incentive to rush a refund the way they have an incentive to rush a charge.

The European Union already solved this. PSD2 — the EU's Payment Services Directive — requires that once a merchant initiates a refund, it must be credited to the customer's account by end of business the following day. Not 3-5 business days. Not up to 14. The next business day. EU banks and payment processors comply with this standard today, using the same card network infrastructure that operates everywhere else in the world. The technology isn't different in Europe — the rule is different.

This platform's position: a card refund — credit or debit, any network — must post to the cardholder's account no later than the next business day after the merchant initiates it. The three-step pipeline (merchant → card network → issuing bank) still exists; the standard requires each party in the chain to treat refund routing with the same priority as charge authorization, not as a lower-priority background task. Merchants have up to their normal return-processing window to decide whether to issue the refund; once they do, the clock starts, and the next business day is when it posts.

Enforcement sits with the CFPB under its existing Dodd-Frank authority over payment systems — no new regulatory body needed, the same agency already overseeing card network rules. Card networks that fail to meet the standard face the same penalty structure already applicable to other payment system violations.