Created by ChatGPT Dec. 2025

By guest blogger Kieran McCarthy

If you have ever wondered why big incumbents keep running to the Northern District of Texas the moment someone builds a tool that makes switching easier, comparing prices easier, or generally makes the internet work like the internet, AT&T Services, Inc. v. T-Mobile US Inc. should help you understand why.

On December 18, 2025, Judge Karen Gren Scholer entered a temporary restraining order blocking T-Mobile from implementing the original “Easy Switch” feature in its T-Life app, and blocking “any substantially similar version” that “accesses or obtains” information from AT&T’s “protected computer systems,” unless T-Mobile gets permission of the Court.

This opinion builds on the case law the N.D. Tex has been generating for years in the Southwest Airlines “don’t you dare build a useful layer on top of our website” cases.

The facts of the case are pretty simple. T-Mobile marketed a feature that let customers log into their current carrier account (AT&T or Verizon) and pull information to help them compare plans and switch. Customers made the decision to switch, and T-Mobile, for obvious reasons, automated the process. The horror!

AT&T sued, saying this was not “customer convenience.” It was unauthorized automated access and scraping of data from password-protected AT&T pages, with allegations of repeated bypassing of AT&T’s blocks, and “over 100” fields of customer data per user.

Over 100 fields? Dang! That’s, like, so many fields! And I suppose there are a few different lenses through which one could analyze that fact. One approach might be to say that automating a process with over 100 fields might be precisely the kind of thing that makes the internet useful, and that saves everyone time, money, and mental headaches.

Another way to view this fact is as evidence of “soooooo much computer fraud” even when T-Mobile is simply automating a process that consumers are choosing to automate. But that is how things work in the Northern District of Texas.

By the time the TRO issued, T-Mobile had already changed the tool so AT&T and Verizon customers could upload a bill PDF or manually enter information.

The court found AT&T likely to succeed because the Easy Switch tool, and its iterations before the November 26 change to PDF upload, accessed AT&T’s systems without authorization, pulled “over 100 fields” of customer data, and transmitted the data back to T-Mobile. It also found irreparable harm to AT&T’s control over its systems and data, plus reputation, goodwill and ‘customer privacy,’ without any inclination to grapple with the awkward fact that the customers were the ones asking to move their own information around. How the court concluded that customer privacy was at issue when the customers themselves initiated the switch 🤷‍♂️.

Even though T-Mobile deactivated the challenged version, the court found the threat remained because T-Mobile wanted to retain the ability to use something “very similar” later.

T-Mobile is enjoined from implementing the original version or any “substantially similar” version, and “substantially similar” is defined basically as anything that accesses or obtains information from AT&T’s protected systems.

Perhaps, learned reader, you might be wondering if there was any discussion of user empowerment, lower lock-in costs, increased innovation and competition, added product development, interoperability, improved price discovery, or any other known policy benefits associated with data portability in the policy section of the TRO?

No. There was not. This is the entire policy discussion of the opinion: “This temporary restraining order will serve the public interest. The enforcement of state and federal laws serves the public interest.”

See how easy this judging stuff is?

To be clear, this is not a case where you would expect someone like T-Mobile to prevail in Texas. But the lack of analysis or consideration for the broader issues at stake is always a bit startling. A big incumbent takes a dispute that is at least partly about competition and consumer switching, recasts it as “computer trespass,” and asks a court to shut the product down quickly. And the N.D. Texas always obliges, especially when the plaintiff is a household-name company with a website and Terms of Use, and the defendant is building a tool that rides on top of it.

That posture matters historically because it reflects an early willingness to treat “automation + Terms + notice” as a path into computer-access liability, even when what is being accessed is, functionally, consumer-facing information.

AT&T’s complaint is explicit that this case is “not about competition for customers,” but about “unauthorized” intrusion into its systems, using automated bots “disguised as an AT&T customer,” scraping “over 100 categories” of data, and bypassing AT&T’s security measures. And that is what is what I like to call “bullshit.”

Either way, the N.D. Texas proves once again why it is the preferred venue and forum for those looking to build walled gardens.

* * *

Interestingly, Texas does have a mandatory data portability law, the TDPSA, or the Texas Data Privacy and Security Act. But the reality is that these laws have very little utility for consumers.

A portability right on paper like the TDPSA is little more than a slow and functionally useless export option. The reality is that laws like this don’t help consumers move with their data.

For one, TDPSA only mandates that companies return the “data you provided,” not the data you actually need to switch. Second, the time, frequency, and authentication friction make it useless for “I want to switch today.” Under TDPSA, controllers generally have up to 45 days (plus a possible 45-day extension) to respond. Waiting 45-90 days for data is so unhelpful that most consumers don’t see any value in requesting it. Next, a .pdf copy of data does not equal “interoperable.” Without shared schemas, APIs, and validation rules, the receiving service cannot reliably ingest the data—and certainly not at scale.

In a case like this one, the consumer-facing promise is “we’ll read your bill and account and recommend the right plan fast.” A statutory portability right typically gives you a dataset, not the transformation, normalization, and comparison workflow that makes switching easy. And when a competitor tries to fill that gap by automating access into the incumbent’s systems, you collide with the CFAA, terms of service, and state computer access statutes (exactly what the TRO discusses). Which is why, without meaningful analysis of the real value of automation for consumers in cases like this one, mandatory portability statutes are functionally useless for consumers.

[Eric’s comment: data portability mandates are generally quite popular, at least in academic circles. But I haven’t seen any evidence indicating that the mandates actually improve anything for anyone. I welcome pointers to academic studies on this topic.]