UPI Charges: What India's New Payment Fee Law Actually Changed

You are at the counter of a small electronics shop in Chennai, phone open, QR code on the screen. The owner waves you off. "UPI has charges now, sir. Cash or card." He is wrong. But he read the same headline you did, he has no way to check it, and by the time this argument ends one of you will have paid for it. Multiply that counter by a few million and you have the actual cost of a law that, on paper, changed nothing about what you owe.

Parliament amended the law that kept UPI free, but it did not impose a single rupee of charge. The amendment moves the decision from statute to government notification. Until that notification exists, nothing changes for you, and person-to-person transfers were never part of the discussion.

Why It Matters

The rule that made UPI free was never a policy of goodwill. It was Section 10A of the Payment and Settlement Systems Act, 2007, and it worked by pointing at another law: the electronic payment modes prescribed under Section 269SU of the Income-tax Act could not be charged for, by banks or by payment providers. Two statutes, chained together. Break the chain and the protection goes with it.

That is precisely what the Taxation and Other Laws (Amendment) Bill, 2026 did. LiveLaw reported that the Rajya Sabha cleared the bill on 11 August 2026, after the Lok Sabha passed it earlier the same month. The amended Section 10A no longer points at the Income-tax Act. It now protects "one or more electronic modes of payment as the Central Government may, by notification, specify." Same section number, same prohibition on charges, completely different source of authority. Protection that used to be automatic is now something a ministry has to grant, and can therefore also decline to grant, without going back to Parliament.

The Finance Minister's office put out a statement calling the change an enabling provision that "does not impose any tax or transaction charge on UPI users." That is accurate and it is also incomplete, which is why both the panic and the reassurance are being read badly. The panic is wrong because no charge exists: no notification has been issued, and a bill that enables is not a bill that levies. The reassurance is wrong if you treat it as permanent, because the whole point of the amendment is to make the next change an executive decision rather than a parliamentary one. If you have followed how the RBI's mis-selling rules moved from draft to enforceable, you already know the shape of this: the gap between "the government may" and "the government has" is where all the real detail sits.

Ask why the change was wanted at all and the answer is duller than the headlines. Every UPI payment consumes something real: switching capacity, settlement, fraud screening, chargeback handling, a support desk when a transfer hangs. Banks and payment providers carry that, and they have argued for years that the government incentive scheme meant to reimburse them for low-value transactions covers only a slice of what the system costs to run. Whether you find that argument sympathetic probably depends on how you feel about bank profitability in general. But it is the argument, and a statute that hard-coded free forever left no legal route to answer it. The amendment opens a route. It does not walk down one.

Numbers explain why the industry pushed for this at all. Free is not free; somebody has been absorbing the switching, settlement and fraud-handling cost of every one of those transactions, and at India's scale that bill stopped being trivial years ago.

Charge on your UPI payment today

₹0

Nothing has been notified

Years UPI ran under the free-by-law rule

6 years

Built on a borrowed clause

UPI value moved in FY2026

₹314 lakh crore

NPCI tally, reported August

Ceiling on debit card MDR, for scale

0.9%

A card rate, not a UPI rate

The debit card ceiling is the one worth sitting with, because it is the comparison every trade proposal is arguing against rather than towards. Nobody serious is suggesting UPI should be priced like a card. The figures being floated in industry submissions sit an order of magnitude below that, and they are aimed at merchants of a certain size, not at the person tapping the phone. The card number is here to show you the ceiling that already exists on a payment method you use without complaint, and to make the point that a fee structure is not automatically a raid on your wallet. It is what gets bundled around a fee later that usually costs you, not the fee itself.

"

The charge on your UPI payment today is still zero. That is a fact about the present, not a promise about next year, and the amendment is what makes the difference matter.

Because the reporting has blurred law, proposal and rumour into one story, it helps to separate them by hand. The table below splits what the statute now says from what is only being discussed in industry submissions and press coverage. Anything in the "reported proposal" column can change tomorrow without a vote.

Category	Detail	What it means
Law	Section 10A now protects modes the Centre notifies, not modes fixed by the Income-tax Act	Power moved from statute to notification
Not law	The amendment levies nothing; the ministry called it an enabling provision	No fee exists until a notification lands
Who pays	Reported proposals apply only to merchants above a turnover line	Your side of the transaction stays untouched
P2P	Person-to-person transfers appear in no proposal reported so far	Sending money home is not in scope
Rates	Industry submissions discuss 5 to 7 basis points, with a per-payment floor around ₹2,000	Discussed, never notified, easily revised
Threshold	Reported figures run from ₹1 crore of turnover at one end to ₹50 crore at the other	That spread proves nothing is settled
Trigger	A gazette notification naming the protected payment modes	Watch that document, not the headlines

Read the threshold row twice. A proposal band running from one crore to fifty crore of annual turnover is not a policy, it is a negotiation still in progress, and the difference between those two numbers is the difference between your neighbourhood chemist paying and only large chains paying. Anyone telling you today which side of that line a shop falls on is guessing. The rate band has the same problem in miniature: the figures being quoted sound precise, but they come from a submission, not from a notification, and submissions get revised for a living.

Four stages, and the only one that can put a fee on a real payment is the last, which has not happened yet.

Friction Points

The genuine risk here is not the statute. It is the twelve months of confusion the statute has just created at street level, where nobody reads gazette notifications and everybody reads forwarded messages. A shopkeeper who believes UPI now costs him money has three options: absorb an imaginary cost, refuse the payment method, or add a line to your bill. Two of those three hurt you, and neither requires any rule to actually exist.

Here is where I will take a position that is not especially popular in the commentary: I think the merchant-versus-consumer distinction is weaker in practice than the official framing suggests. Not because the rule is dishonest, but because a merchant fee that is real gets priced into what the merchant sells, the same way rent and electricity do. A fee at the scale being discussed is small enough that this may never show up on a price tag. That is a reasonable expectation, and it is still only an expectation. Nobody can point to evidence either way yet, because no such fee has ever run on UPI at national scale.

The redress path is the other soft spot. If a shop does start adding a surcharge, your complaint goes to the acquiring bank, the payment app, or the shop itself, and anyone who has fought a large service provider's automated support queue knows how much energy that costs for a twelve rupee dispute. Most people will pay and move on. That, quietly, is how a rule that applies to nobody becomes a cost that lands on everybody.

• Treat any message claiming a specific UPI charge percentage as false until you can find the gazette notification behind it.
• A shop that adds a UPI surcharge to your bill is making a commercial decision, not following a new law. Ask which rule it is following.
• Screenshot the bill line and the payment confirmation before you leave. A dispute raised the next day without both is close to unwinnable.
• Watch your payment app's terms-of-service update mails rather than news alerts. Fee changes appear there first, and the same habit that catches a quietly changed filing form catches this.

Key takeaways

Four things worth carrying out of this, none of which need a lawyer:

• Nothing at the counter has changed. The amendment altered where the authority sits, not what you pay.
• The trigger to watch is a gazette notification listing protected payment modes. Until it exists, every rate you read is somebody's proposal.
• Person-to-person transfers have stayed outside every version of this discussion so far.
• If a surcharge does appear on your bill, ask for it in writing. A shop that will not itemise it has told you what it is.

So do one thing this week. The next time somebody tells you UPI now has charges, ask them for the notification number. They will not have one, because it does not exist, and that single question is the entire defence you need for as long as that stays true. Set a reminder to ask it again in six months, when it might not.

RBI's New Mis-Selling Rules: What Bank Customers Can Now Demand

The loan officer slides three forms across the desk. One is the home loan you actually came for. The other two are a life cover and a monthly investment plan, already filled in, already ticked, and he mentions that the file tends to move faster this way. You sign all three, because the sanction is two days out and arguing feels expensive. That exact scene, repeated a few million times a year across Indian branches, is what the Reserve Bank spent this February trying to make indefensible.

India's banks can no longer bundle an insurance policy onto your loan, tick consent boxes on your behalf, or let agents pass themselves off as staff. If a sale is judged unsuitable for your profile, RBI's draft rules make the bank refund every rupee.

Why Your Signature Stopped Being the Bank's Defence

The interesting move here is not the list of bans. It is the definition. RBI's draft Amendment Directions, issued on 11 February 2026 after the policy statement of 6 February, define mis-selling to cover the "sale of a product / service, which is neither suitable nor appropriate in view of the customer's profile even if with his / her explicit consent." Read that last clause again. For years the branch's answer to any complaint was a signed form with your name on it. Under this draft, that form settles nothing.

Suitability is not left as a mood, either. A bank has to map what it sells against your age, income, employment profile, financial literacy, risk tolerance and investment horizon, classify its own products by complexity and risk, and write down the logic connecting the two. Most coverage treated this as compliance paperwork. That reading undersells it badly. Suitability is the load-bearing wall of the whole framework: bans on specific tricks age badly, because sales teams invent new ones, but a written suitability test follows the product wherever it goes. The forced-package problem is not unique to banking, and readers who followed our breakdown of the scam behind DTH combo channel packages will recognise the shape of it instantly.

Consent gets rebuilt from the ground up too. It has to be specific, informed, unambiguous and captured by a clear affirmative action, and it is required before the bank even approaches you, not merely before you sign. Having a loan with a bank no longer entitles that bank to pitch you insurance. Pre-ticked boxes and catch-all lines about receiving offers from partners are out. Anyone who has tried to trace who exactly sold their number to a call centre will find this familiar territory, much like the gaps we covered in TRAI's proposal for universal caller ID. There is a sharper provision buried in the staff rules: employees who market third-party products cannot accept any direct or indirect incentive from the company whose product they are pushing. That single line does more damage to aggressive branch selling than the entire consent chapter.

None of this arrives in a vacuum. Complaint volumes are what forced the regulator's hand, and they explain why the draft reaches for compensation rather than a warning letter. The figures below set the scale of the problem and the size of the remedy now attached to it.

COMPLAINT WINDOW

30 days

From signed terms received

REFUND OWED

100%

Entire amount you paid

OMBUDSMAN LOAD

13.3 lakh

RB-IOS complaints in FY25

YEAR-ON-YEAR RISE

13.55%

Growth over FY24 volume

The complaint window is the number to write on your fridge. It does not start when you notice the problem. It starts when the signed copy of your terms lands with you, which for most people is an email attachment they never open. Miss it and the refund route defined in the draft closes, leaving you back at the ordinary grievance queue, where the clock is longer and the outcome is thinner. Read the attachment the day it arrives.

Branch Practice Before, and What the Draft Demands

Most of these obligations only make sense when you set them beside what a branch actually does today. The left column is not a caricature. It is standard practice at a large number of Indian bank counters, and every line of it is now addressed by a specific clause.

Practice Area	Common Branch Practice	What the Draft Requires
Consent	One blanket clause covering the bank and all its partners	A separate affirmative opt-in per product and per purpose
Bundling	Loan sanction quietly conditional on taking the cover	Compulsory bundling prohibited outright
Application forms	One combined form, add-ons pre-filled by staff	A distinct form per product, naming the product type up front
Who is selling	Agents at a desk inside the branch, indistinguishable from staff	Visible ID, badge or desk signage marking them as agents
Price via an agent	Never mentioned, often higher than going direct	Any rate or fee difference must be disclosed to you
App and web design	Decline buried, accept glowing, repeat pop-ups after refusal	Dark patterns banned, interfaces user-tested and audited
After you apply	Silence until the first premium debit shows up	An SMS or email asking you to confirm you applied at all
Best suited for	Branches chasing quarterly fee income targets	Customers who want a paper trail they can act on

Look down the right column and a pattern shows up. Almost every clause creates a record: a timestamped consent, a separate form, a confirmation message, an audit of the app screen. That is deliberate. A rule you cannot prove was broken is a rule nobody enforces, and this draft is mostly an exercise in making the bank generate the evidence against itself. The dates below show how fast the regulator moved once it decided to act.

Five days from announcement to published draft, and roughly five months from there to the expected commencement date.

Where This Gets Messy

The agent rules are the most concrete thing in the package and the easiest to check. Direct selling agents may contact you only between 09:00 and 18:00 hours unless you have expressly authorised otherwise, they cannot turn up at your home or workplace without your explicit consent, and they must hand over their supervisor's details if you ask. That is a real, enforceable boundary. Whether anyone answers the phone when you report a breach is a separate question, and the honest answer is that we do not know yet. Our account of Airtel's automated support response is a fair map of how long that road can run once a large company decides your complaint is a ticket rather than a problem.

Then there is the grey area nobody has resolved. Compulsory bundling is defined as making one product conditional on another, but it carves out packages offered free of extra cost. So what about the savings account you are told to open before the personal loan can be processed? It costs you nothing directly. It also is not optional. The draft can be read both ways, and until the final text or a clarification lands, that ambiguity is worth exactly as much as the branch manager wants it to be worth. Paperwork ambiguity has a long history of landing on the customer's side of the counter in this country, as anyone who has fought through India's tax filing forms and process will tell you.

A few things to keep in view before you assume the problem is solved.

• This is still a draft. It went out for stakeholder comment, and the text that commences may differ from the version being analysed today.
• The refund route runs through the bank's own approved policy first, so the compensation you get for a proven bank mis-selling case depends on a document the bank wrote.
• Consent given in the past does not disappear. If you ticked a blanket box three years ago, expect to have to withdraw it actively rather than watch it lapse.
• Enforcement lives with branch-level supervision, which is precisely where every previous fair-practice code went to die.

FOUR RIGHTS WORTH REMEMBERING

▸ Consent is needed before the bank may even approach you about a product, not just before you sign for it.

▸ Funding an add-on from your loan amount needs its own separate consent, distinct from agreeing to buy it.

▸ Unsubscribing must be as easy as subscribing, with everything you are signed up for listed on your login page.

▸ The post-sale feedback call has to come from a team with no connection to the sale itself.

Do one thing this week. Log into your bank's app, open the consents or communication preferences page, and look at what you have apparently agreed to. Most people find at least one blanket permission they do not remember granting, and the draft rules give you a reason to strip it out now rather than after the next call from a number you do not recognise. Then find the signed terms of anything a branch sold you in the last month and read them properly. The clock on bank mis-selling redress is short, and it is already running.

Robot Vacuum Buying In 2026: Why Mid-Range Beats Expensive Flagships

Your neighbor just parked a fifteen-hundred-dollar cleaning robot in the hallway, docking tower and all. It laser-maps every room, scrubs and dries its own mop pad, and empties itself for two months without a single thought from anyone. It also cost more than a used motorcycle. Down the street, a sub-$200 disc grabs the same dog hair every morning, and nobody can tell the two floors apart. That gap — between the sticker and what actually lands in the bin — is where 2026 shoppers get quietly fleeced.

TL;DR: In 2026 a $400–$500 mid-range robot vacuum handles the same daily mess as the priciest flagship. Pay for LiDAR navigation and a self-emptying dock, skip the ultra-premium mop-scrubbing towers, and keep a corded upright around for the deep carpet cleans no disc can manage.

Why Cheap Discs Suddenly Clean Like Expensive Ones

The one thing that used to separate a smart floor robot from a dumb one was navigation. A machine that bounces off the walls at random is like mowing a lawn blindfolded — it covers the ground eventually, just badly, missing strips and re-cleaning others while the battery drains. LiDAR fixed that by handing the robot a laser map of your rooms. And as of Q2 2026, that mapping hardware, paired with a base that empties the bin for you, shows up on models under $300 — gear that sat firmly in premium territory a year earlier, per Vacuum Wars' 2026 rankings.

So what does the extra thousand dollars actually buy? Mostly mopping theatrics. Self-washing mop pads, warm-air drying so they don't sour, an auto water-refill tank, and RGB cameras that swerve around a stray sock. Real conveniences, every one. But none of them change whether your floors are clean by Friday. And that's the sleight of hand: the premium tier sells you a maintenance-reduction story, not a cleaning-quality one.

The daily tax of the wrong pick stays invisible at checkout and shows up later as chores. Emptying a dockless bin runs about 41 seconds a pass; twice a day with a shedding dog, that is roughly five hours a year spent hunched over a trash can. A self-emptying base erases that entirely. The same money logic drove our breakdown of why DIY SSDs outshine overpriced external drives — the premium badge rarely buys performance in the same proportion as it buys price. The numbers below sketch the real shape of ownership.

Service Life
4–6 yrs
Before a motor swap

Consumable Cost
$25
Per mop-pad refill pair

Budget Suction
12,000 Pa
On sub-$230 LiDAR units

Value Retention
90%
Of flagship cleaning ability

Take the running-cost figure. The consumables look trivial line by line — a mop-pad pair here, a filter there, a dust bag every couple of months. But a scrubbing flagship keeps feeding all of them on a schedule a mopless budget disc simply never triggers, because it has no mop to wash and no reservoir to refill. The cheap machine's biggest saving isn't the sticker. It's everything it quietly never asks you to buy again.

The 2026 Shelf, Ranked By What Matters

Line the shelf up by the things you actually feel — can it map, can it empty itself, will it lift fur off a rug — and the price tiers stop looking like a quality ladder. They look like a convenience ladder, where the bottom rung already cleans. Here is how six representative 2026 models stack against each other.

Model	Price	Suction	Navigation	Hands-Free Dock	Mopping	Obstacle Avoidance	Best Suited For
Roborock Q5 Max+	$160	5,500 Pa	LiDAR	Self-empty bag	None	Basic	Bargain hunters, hard floors
DREAME D10 Plus Gen 2	$179	6,000 Pa	LiDAR	90-day auto-empty	Light	Basic	First-timers wanting long dock gaps
Shark AI Ultra	$250	7,200 Pa	360° LiDAR	30-day bin	None	Basic	Big US homes, brand-name buyers
eufy X10 Pro Omni	$450	8,000 Pa	LiDAR	Self-wash + dry mop	Full, self-washing	Camera-assisted	Most people, mixed floors
Roborock Qrevo 35A	$460	7,000 Pa	LiDAR	Empty, wash, dry, refill	Full, auto refill	Camera-assisted	Hands-off households, pets
Dreame X60 Max Ultra	$820	20,000 Pa	Retractable LiDAR + AI	Full omni station	Full, warm-water wash	Best-in-class	Thick-carpet homes, gadget lovers

Read down the navigation column. Every single row says LiDAR. The thing that used to justify the price jump is now table stakes, and the money above $450 buys a fancier dock, not a cleaner floor. Only the top row's suction is genuinely in another league, and that matters to exactly one buyer: the one fighting embedded fur in thick carpet. Everyone else is paying for the tower, not the clean.

A realistic ownership rhythm: heavy setup once, near-zero attention daily, and a short seasonal chore to swap the sealed dust bag.

The Catch Nobody Prints On The Box

Carpet is where the marketing quietly falls apart. A disc that glides across hardwood can wheeze on a medium-pile rug, because lifting fur out of woven fibers needs real suction and a stiff brush, not a spec-sheet number. Here is the honest grey area: obstacle avoidance still has no clean answer. Structured-light sensors are cheap but plow straight into cables and socks; RGB cameras spot the clutter but cost more and put a lens in your living room. Neither reliably clears a floor a toddler just decorated. Anyone promising a robot that never eats a charging cable is guessing.

The upkeep is real, too, and it rewards the same habit that keeps any gadget alive — the one we pushed in our look at smarter MacBook battery management: small, regular attention beats heroic rescue. Hair still wraps the roller. Filters still clog. A dock that scrubs its own mop still needs its tray rinsed now and then, or the whole machine starts to smell like a wet dog.

• Match suction to your floors — hard surfaces are forgiving, but plush carpet needs a genuinely strong motor, not a marketing figure.
• Do a thirty-second floor sweep before each run if you own a budget model without a camera, or plan to fish a sock out of the roller.
• Price the consumables before you buy: bags, filters, brushes and pads are the real cost of ownership, not the one-time sticker.
• Skip the ultra-high suction hype unless embedded pet hair on thick carpet is your actual daily problem.

Key Takeaways Index Card

➤ In a 2026 test of 23 models, only 2 dodged more than 9 in 10 floor obstacles — and both were expensive.
➤ 3,000 Pa is the realistic floor for pulling fur out of medium-pile carpet; anything less just grooms the surface.
➤ A dockless budget unit with a heavy shedder can beg to be emptied 2–3 times a day.
➤ Run the robot 3–4 times a week and the corded upright still owns the quarterly deep clean no disc replaces.

Pick the cheapest model that has LiDAR and a self-emptying dock, spend the saved money on a good corded upright for the deep cleans, and only climb into flagship pricing if you are genuinely losing the war against pet hair in thick carpet. The premium tower is buying you a chore you could have skipped for a few hundred dollars less.