Cloning AI is not a not a strategy, it's a lawsuit.

Something remarkable happened in the startup world in March 2026, the kind of story that makes IP attorneys reach for their highlighters.

Todd Saunders, a SaaS veteran and former Google executive who built Broadlume into a leading vertical SaaS platform before its acquisition, posted something on X that went immediately viral. He described a scenario he’d personally heard about: a seed-stage startup, backed by a well-known venture capital firm, had put in its actual board deck a go-to-market strategy built around gaining access to a large incumbent software product as a paying customer, using AI coding tools to clone the entire product, and offering it to the market at 90% less than the original. The VC reportedly endorsed the plan in writing.

Paul Roetzer and Mike Kaput of The Artificial Intelligence Show flagged it in Episode 202 as a vivid example of the “SaaS apocalypse” dynamic reshaping the industry, where tools like Claude Code have made it technically trivial to clone software that once required years and millions of dollars to build. Saunders called the strategy “ethically bankrupt.”

We’d call it something more specific: a multi-front legal catastrophe hiding inside a pitch deck.

Let’s Start With the Terms of Service, Because That’s Where It Ends.

The moment this startup signed up as a paying customer, they entered a binding legal agreement. Every commercial SaaS product comes with a Terms of Service or End User License Agreement, and nearly every one of them contains provisions flatly incompatible with what this startup planned to do.

Standard prohibitions in SaaS agreements include: no reverse engineering or decompiling of the product’s underlying architecture; no use of the platform for competitive purposes or to build substitute products; no automated scraping or benchmarking for competitive intelligence; and a license restricted to the subscriber’s internal business use only. Using a licensed subscription to systematically study, map, and replicate the platform, with documented intent to build a commercial clone, almost certainly constitutes breach from day one.

One important nuance: a ToS is only as enforceable as its acceptance flow. Courts increasingly distinguish between browse-wrap agreements (terms buried in a footer link, often unenforceable) and click-wrap flows that require affirmative acceptance. Clear conspicuous links and documented acceptance are essential for a ToS to hold up in litigation. That said, enterprise SaaS products almost universally use click-wrap agreements, so a paying customer almost certainly accepted enforceable terms.

The remedies for breach can include license termination, disgorgement of profits, compensatory damages, injunctive relief, and attorney’s fees. For a seed-stage startup, a well-funded plaintiff’s legal response can be existential before a product ever ships.

Then There’s the Trade Secret Problem

Layer in trade secret law, and the exposure deepens considerably.

Under the Defend Trade Secrets Act, software architecture, proprietary algorithms, UX logic, data models, and feature implementations can all qualify as protectable trade secrets. When a company uses its position as a licensed customer to extract and replicate those elements, particularly when that intent is documented in a board deck, the conduct shifts from breach of contract into federal misappropriation territory.

The DTSA defines misappropriation to include acquiring a trade secret by “improper means.” Courts have interpreted this broadly: access obtained through deception, or used far beyond the scope of what was authorized, can qualify. Here, the argument writes itself. The startup obtained a license representing itself as a legitimate customer, then used that access as a vehicle for competitive intelligence and cloning. That’s not a gray area.

Under the DTSA, successful plaintiffs can recover actual damages, the defendant’s unjust enrichment, injunctive relief, and, for willful and malicious misappropriation, exemplary damages of up to twice actual damages, plus attorney’s fees. For a startup that hasn’t yet shipped a product, an injunction alone can be fatal.

Which brings us to the single most damaging fact in this story: the strategy was in a board deck. In writing. Endorsed by the VC in writing. That’s not just evidence of misappropriation, it’s evidence of willful misappropriation. In trade secret litigation, that distinction is the accelerant that turns a serious lawsuit into a catastrophic one.

The “We Used AI” Defense Doesn’t Work

Some founders in this position instinctively argue that since an AI tool did the actual coding, the output is their own independent creation. That argument misunderstands how trade secret law operates.

What matters is not how the information was taken, it’s that protected information was used without authorization. If a founder studied a competitor’s product through licensed access, fed observations and structural insights into an AI tool, and directed that tool to reconstruct what they observed, the AI is simply the instrument of the misappropriation, not a shield from it.

The Eleventh Circuit reinforced this principle in Compulife Software v. Newman (August 2024), holding that competitors who used automated bots to scrape millions of insurance quotes from a publicly accessible website had acquired trade secrets by “improper means”, even though the underlying data was technically visible to the public. The method of acquisition and the intent behind it mattered as much as the access itself. AI-assisted product cloning presents the same legal logic: the tool changes, the analysis doesn’t.

A December 2025 analysis by IPWatchdog put it directly: AI has made reverse engineering dramatically faster, and SaaS platforms are a specific emerging target. The analysis flagged AI-assisted cloning as a growing legal threat and noted that the infrastructure to pursue such claims is already in place, published months before this story broke.

There may also be a copyright dimension. If the clone reproduces the structure, sequence, or organization of the original software’s code, copyright infringement claims under the Computer Associates v. Altai framework may apply alongside trade secret claims, compounding exposure further.

The Investor Exposure Is Real Too

The detail most observers aren’t fully processing: the VC who endorsed this strategy in writing carries real legal exposure of their own. Investors who knowingly encourage conduct constituting trade secret misappropriation or breach of contract can face liability through contribution and indemnification theories, and potentially tortious interference claims if their involvement in directing the infringing strategy rises to active participation. Venture firms are not insulated from the conduct of their portfolio companies when documentary evidence shows awareness and endorsement of the specific wrongdoing.

This is precisely why experienced investors and their counsel are careful about what they commit to writing when advising on competitive strategy.

AI Changed the Cost of Copying. Not the Law Against It.

AI coding tools have made it technically trivial to clone software that once required years of development. But the legal framework governing what you can build has not moved an inch. The speed of copying doesn’t change the wrongfulness of copying. AI doing the work creates no new exception in trade secret law. And a VC’s written endorsement doesn’t convert a breach of contract into a business model.

If anything, AI has made the legal risk more acute. The faster and cheaper cloning becomes, the more aggressively incumbent software companies will monitor for misuse of licensed access, and litigate when they find it. SaaS agreements are already being updated to explicitly prohibit AI-assisted reverse engineering and competitive benchmarking. The legal net is tightening at precisely the moment the technical barriers are falling.

The same AI tools that make cloning possible also make it faster than ever to build something genuinely original and defensible. Founders who channel that capability into novel development, and protect it with real IP strategy from day one, are positioned for durable competitive advantage. The ones who treat AI as a shortcut to copying will learn, at the worst possible time, that the law hasn’t kept up with their assumptions.

Building Something? Let’s Make Sure You Own It.

At Griffith Barbee, our Dallas IP attorneys represent businesses on both sides of trade secret and software IP disputes, and we help founders, SaaS companies, and investors build protection strategies that hold up before a lawsuit arrives. If you’re navigating competitive strategy in the AI era, we’d rather help you get it right upfront than meet you after the complaint has been filed.

📞 Contact us today to schedule a consultation. Smart IP strategy is a competitive advantage, and in the AI era, it matters more than ever.