The SAFT, and why its legal standing is still argued about.

What the SAFT was designed to do

In October 2017, at the height of the initial coin offering period, Cooley LLP and Protocol Labs published "The SAFT Project: Toward a Compliant Token Sale Framework". The authors were attempting to solve a real problem. Teams were selling tokens to the public to fund the development of networks that did not yet exist, and doing so in a way that looked, to many observers and eventually to regulators, like an unregistered public offering of securities.

The SAFT proposed a structure with two distinct moments. First, an accredited investor signs an agreement and pays money, receiving a contractual right to tokens in the future. The drafters accepted that this agreement is a security and should be sold under a private placement exemption. Second, when the network is live and the token has genuine functional utility, the tokens are delivered. The argument was that the delivered token — now a consumptive good used to access a working network — was a different thing from the investment contract that funded its creation.

The intellectual move is the separation. If it holds, a team can raise development capital from professional investors under a well-understood exemption without the eventual token carrying securities status forever. If it does not hold, the two steps collapse into a single offering, and the exemption relied on for the first step may not survive contact with the second.

The underlying legal test is the one set out by the Supreme Court in SEC v. W.J. Howey Co., 328 U.S. 293 (1946): an investment contract exists where there is an investment of money in a common enterprise with a reasonable expectation of profits to be derived from the efforts of others. Every argument in this area, including the SAFT’s, is ultimately an argument about how Howey applies to a particular set of facts.

SAFT and SAFE are not the same instrument

The naming similarity causes real confusion, including among founders who assume a SAFT is simply the crypto version of the document they already signed. The mechanics differ in the way that matters most: what the investor ends up holding.

Why the SAFT’s standing became contested

Criticism did not wait for litigation. Within weeks of the whitepaper, the Cardozo Blockchain Project published "Not So Fast — Risks Related to the Use of a ‘SAFT’ for Token Sales", arguing that the structure carried substantial risk and that its central separation might not hold under existing law. That report is worth reading alongside the original whitepaper, because it frames the dispute before any court had spoken.

The first substantial judicial treatment came in SEC v. Telegram Group Inc. in the Southern District of New York. On 24 March 2020, Judge Castel granted a preliminary injunction preventing Telegram from distributing its Gram tokens. The court’s reasoning went to the heart of the SAFT premise: it declined to analyse the purchase agreements and the Grams as separate objects, and instead considered the scheme as a whole, in which initial purchasers were expected to distribute Grams into a secondary market. On that view the separation the structure depended on did not exist on those facts.

Later that year, on 30 September 2020, Judge Hellerstein granted the SEC summary judgment in SEC v. Kik Interactive Inc., holding that Kik’s offering of Kin was an unregistered offer and sale of securities. The court treated the earlier pre-sale to accredited investors and the subsequent public token distribution event as a single integrated offering, with the consequence that the exemption relied on for the pre-sale was unavailable. The SEC announced a final judgment including a five million dollar penalty and a three-year notice requirement for certain future digital asset issuances.

The picture became more granular in SEC v. Ripple Labs. On 13 July 2023, Judge Torres held that Ripple’s direct sales of XRP to institutional buyers were offers and sales of investment contracts, while programmatic sales through exchanges on those facts were not — a distinction drawn on what each category of purchaser could reasonably have understood about who they were buying from and what had been promised. The parties dismissed their respective appeals in 2025, so the split ruling stands as a district court decision rather than an appellate rule.

Read together, these cases point at something more useful than a verdict on the SAFT. They suggest the analysis attaches to transactions and the circumstances around them, rather than to a token as a permanent intrinsic property. The same token can be sold in one way that constitutes an investment contract and another way that, on those facts, does not.

Where the analysis stands in 2026

For years the reference point for practitioners was the SEC Division of Corporation Finance’s 2019 "Framework for ‘Investment Contract’ Analysis of Digital Assets", a non-binding analytical tool published by FinHub. That document has been withdrawn. The SEC’s own page for it now marks it as superseded by the interpretation issued on 17 March 2026.

That interpretation — "Application of the Federal Securities Laws to Certain Types of Crypto Assets and Certain Transactions Involving Crypto Assets", Release Nos. 33-11412 and 34-105020, joined by the CFTC — sets out a five-category taxonomy: digital commodities, digital collectibles, digital tools, stablecoins, and tokenised securities. It describes a non-security crypto asset as becoming subject to an investment contract where an issuer’s representations or promises to undertake essential managerial efforts would cause purchasers reasonably to expect profits, and describes it as ceasing to be subject to one where purchasers can no longer reasonably view those representations as remaining attached to the asset. It also addresses airdrops, protocol mining, protocol staking, and wrapping.

What teams tend to use instead

Practice shifted after 2020, though not to a single replacement. What is observable is a move away from standalone token pre-sales towards instruments that sit alongside an equity round, so that the investment is primarily in a company and the token exposure is an attached right rather than the whole deal.

• A priced equity round or SAFE, with a separate token warrant giving the investor a right to purchase tokens if and when they are issued — covered in detail at /equity/token-warrant.
• A SAFE plus a token side letter, where the token entitlement is granted rather than purchased separately, sized by reference to the equity position.
• Token purchase agreements executed after a network is live, with lock-ups and transfer restrictions, rather than pre-launch agreements funding development.
• Structures that route the token through a separate issuing entity, which raises the reconciliation problem described at /equity/dual-cap-table.
• Non-US offerings under local regimes, where the analysis is a different body of law entirely and no US-centric framing applies.

None of these is a safe harbour. They are patterns that have become common because they distribute risk differently, not because any regulator blessed them. The reason they matter for a cap table is that each one creates a different kind of row: a token warrant is a contingent claim on future supply, a side letter is a grant sized off an equity position, and a post-launch purchase agreement is a delivered balance with restrictions. Recording them as if they were interchangeable is how token registers become unreliable.