What happened
On September 1, 2026 the Securities and Exchange Commission proposed to modernise the rules governing registered transfer agents (the firms that keep the official record of who owns a company's shares) for the first substantive time since they were written in the late 1970s and early 1980s. The proposal amends existing rules and forms, rescinds one, and adds new ones, among them written compliance policies and a rule on restrictive legends, which are the notices that bar a share from resale (Decrypt, Blockhead). New questions on Form TA-2 would ask agents how many of their issues keep the master securityholder file (the authoritative ownership list) on a distributed ledger, separating issuer-sponsored from third-party-sponsored tokenized securities. Comments run 60 days from Federal Register publication.
The framework read
Two documents describe this event. The crypto press ran it as a blockchain overhaul. The Commission's own release is headlined "Modernize Rules for Registered Transfer Agents," and names blockchain once, in a list that starts with electronic communications: Chairman Paul Atkins said the rules would reflect agents' "current processes and operations, including the use of electronic communications and blockchain technology." The gap between them is the thing worth a reader's attention.
Technology-neutrality is a decision here, not an omission. Blockhead reads the release as stating that the proposal does not "approve any specific blockchain-based transfer system or grant blockchain-based transfers different regulatory treatment than paper or electronic ones." The Form TA-2 change is a census question (how many, sponsored by whom) rather than a new category. The Commission is proposing to regulate the job, and to stop caring what the record is written on.
A rules-based rating framework makes the same choice, for the same reason. Nothing in the six weighted dimensions scores what a token runs on. There is no ledger-quality input, no bonus for being on-chain, and no gate a technology passes or fails.
Which is why a tokenized share is not a crypto asset, and the framework has almost nothing to say about one. Tokenomics & Value Accrual is the heaviest dimension at 28 of 100 points and reads supply, emissions, unlock cliffs and float against fully diluted valuation. A tokenised share has none of those: its supply is a share count set by a board, and there is no emissions schedule to read. On-chain Usage & Economics, 22 points, reads a network's own fees, users and revenue, not the revenue of a company whose stock happens to settle there. Security, Decentralization & Durability, 18 points, reads node count, client diversity, governance and audit quality; a tokenized share's equivalent risk is the transfer agent's cybersecurity and business continuity, precisely what the SEC has proposed to regulate. Same rails, different cargo. This framework rates assets, not registrars.
One line in the proposal can eventually reach a crypto asset's score, and it is a fee line. If regulated securities settle on a public chain, that chain gets paid, and On-chain Usage & Economics is explicitly about real, non-incentivised volume, fees and revenue. That is the only channel: not legitimacy, not sentiment, not an institution's name in a press release, but money booked by the network. Yesterday's post drew the same line about buybacks: the scoreable question was never whether a programme exists, but what pays for it. That number is small today and this proposal does not raise it, as it makes the attempt lawful and adds a form field to count it.
The neutrality cuts the other way too. If a chain does win securities settlement, it earns no credit for being a chain: it gets scored on fees and users like any other business with customers. A rule that refuses to privilege the technology also refuses to reward it.
None of this moves a grade; grades come from the scoring engine applying the framework to verified data, not from a news cycle. Of the 203 assets covered, 161 grade F, average composite 27.3 out of 100 (data as of August 31, 2026), and the recurring failure modes are unchanged: supply concentrated in insider wallets, yield that cannot be paid out of real revenue, liquidity too thin to exit. A token does not become sound because a regulator has learned the word for its database.
What we're watching
Federal Register publication, which starts the 60-day clock. Until it lands there is no deadline.
The Form TA-2 answers, if the rule is adopted. This would produce the first official count of how many securities issues keep a master securityholder file on a distributed ledger, split by who sponsored it. A field that runs on vendor projections would acquire a number somebody had to certify.
Which chains, and what they are paid in. Settlement that pays fees in a network's native asset shows up in that network's revenue. Settlement where a platform bills an issuer in dollars does not touch any token's score, however much volume crosses it.
How restrictive legends survive contact with a public ledger. A share barred from resale needs some party able to stop the transfer. Whether that resolves as a permissioned ledger, a wrapper, or an agent with keys is where "tokenized" gets its working definition.
The comment file. Incumbent transfer agents and tokenization platforms want different things from the same 60 days, and the operational detail surfaces in their disagreement, not in the release.