Regulation, Custody & the Real Risks of Web3

Smart Contracts, DeFi & Web3 Security
Course 2 · Chapter 9 · Regulation, Custody & the Real Risks of Web3

Chapters 7 and 8 covered risk that lives inside a smart contract, or in the infrastructure directly operating one. This chapter steps back further — to a risk category that has nothing to do with code at all, and to the real, still-evolving legal frameworks governments have built in response to it.

Custody: Who Actually Holds Your Assets?

Course 1 covered private keys as the real, sole source of control over an on-chain asset — "not your keys, not your coins" is the community's own well-known shorthand for that principle. A custodial exchange breaks that principle by design: when you deposit crypto onto a centralized exchange, the exchange itself holds the private keys, and your account balance is really just an internal database entry the exchange promises corresponds to real, held assets. A non-custodial wallet, by contrast, keeps you holding the actual private key the entire time.

The Real, Practical Trade-off Custodial exchanges are genuinely more convenient — easier onboarding, password resets, no risk of permanently losing a seed phrase. Non-custodial wallets are genuinely more aligned with Course 1's own trust-minimization theme — nothing can freeze, lose, or misuse your assets except you. Neither option is simply "correct"; each trades real convenience against real, different categories of risk.

Case Study: The FTX Collapse (November 2022)

FTX, founded in May 2019 by Sam Bankman-Fried and Gary Wang, grew into one of the world's largest cryptocurrency exchanges, reaching a real, reported $32 billion valuation before its collapse. The exchange filed for Chapter 11 bankruptcy on 11 November 2022.

2 Nov 2022
A CoinDesk report reveals that Alameda Research, FTX's own affiliated trading firm, held a large share of its balance sheet in FTT, FTX's own native token — a real, circular dependency.
Early Nov 2022
Binance announces plans to sell its own FTT holdings, triggering a real, rapid loss of confidence and a wave of customer withdrawals.
11 Nov 2022
FTX files for Chapter 11 bankruptcy. New CEO John J. Ray III later calls it "a complete failure of corporate controls."
12 Dec 2022
Bankman-Fried is arrested and charged with fraud and money laundering.
2 Nov 2023
A jury convicts Bankman-Fried on all seven real counts, including wire fraud and conspiracy to commit securities fraud.
28 Mar 2024
He is sentenced to 25 years in prison and ordered to forfeit $11.02 billion.

The real, documented allegation at the center of the case: FTX had lent approximately $10 billion in customer deposits to Alameda Research, and Alameda had reportedly been granted a secret exemption from FTX's own automatic liquidation safeguards — the exact kind of privileged-access failure Chapter 7's own access-control category describes, just happening inside a company's internal systems rather than inside deployed Solidity code.

The Real Irony This Case Study Delivers FTX was a fully centralized, custodial company — the exact opposite of the trust-minimized, self-custodied model Course 1 spent its own chapters explaining. Depositing funds onto an exchange like FTX meant trusting a company's own internal controls completely, with none of the on-chain transparency or self-custody guarantees a personal wallet actually provides. The collapse is a real, costly demonstration of exactly the risk "not your keys, not your coins" warns against.

Regulation: The US Approach

US securities regulation still leans heavily on a real, 78-year-old legal test. In SEC v. W. J. Howey Co. (328 U.S. 293, decided 27 May 1946), the Supreme Court held that an investment contract — and therefore a regulated security — exists whenever there is (1) an investment of money, (2) in a common enterprise, (3) with an expectation of profit, (4) derived primarily from the efforts of others. The real, ongoing debate over whether a given token counts as a security largely comes down to whether it satisfies all four of these prongs.

Regulation: The EU's MiCA

The European Union took a genuinely different approach: a dedicated, purpose-built regulation rather than an 80-year-old test stretched to fit a new technology. The Markets in Crypto-Assets regulation (MiCA) was adopted by the European Parliament on 20 April 2023 and applied in full via two real phases — stablecoins (asset-referenced and e-money tokens) from 30 June 2024, and other crypto-assets plus crypto-asset service providers (exchanges, custodial wallet providers) from 30 December 2024.

US (Howey Test)EU (MiCA)
Real originA 1946 Supreme Court ruling about Florida citrus groves, applied by analogyA dedicated 2023 regulation written specifically for crypto-assets
Real mechanismCase-by-case application of a four-part test to determine security statusA direct regulatory framework covering issuers, exchanges, and custodial wallet providers by name
Real effect on this chapter's own FTX caseWould ask whether specific FTX products (like the FTT token) met the Howey criteriaWould have directly regulated FTX's own custodial and exchange functions had it applied at the time

Risks Beyond Code and Custody

Two more real, well-documented Web3-specific risk categories round out this chapter's own survey. A rug pull is when a project's own developers abandon it and drain its liquidity pool (Chapter 4) after attracting real investor funds — a risk that has nothing to do with a smart contract bug at all, since the contract may be working exactly as its own creators always intended it to. A approve() phishing exploit targets a real, genuine ERC-20 feature (Course 1's own token standard) that lets a user grant a smart contract ongoing permission to move tokens on their behalf — a malicious site can trick a user into granting that permission to an attacker-controlled contract, which then drains the approved tokens directly, with no bug in the token contract itself required.

A Genuine Pattern Across This Whole Chapter FTX, rug pulls, and approve() phishing share one real, unifying lesson: none of them require a single line of flawed Solidity code. Chapters 7 and 8 covered risk inside the code; this chapter covers risk around it — who holds your keys, who regulates the platform you're using, and who you trust with a permission you granted. A perfectly secure smart contract cannot protect against any of these on its own.

Hands-On Exercises

Three exercises applying this chapter's own custody, regulation, and risk material before Chapter 10's own capstone brings the full course together.

Exercise 1
Explain, using this chapter's own custodial-vs-non-custodial distinction, exactly why depositing funds onto FTX exposed customers to a real risk that holding the same funds in a personal, non-custodial wallet never would have.
Exercise 2
Apply the real, four-part Howey Test to a hypothetical new token sold with the pitch "buy this token now, and our development team will build products that make it valuable later." Explain which of the four prongs this pitch appears to satisfy, and why.
Exercise 3
Explain why a rug pull and an approve() phishing exploit both fall outside Chapter 7's own smart contract vulnerability taxonomy, even though both can result in a user losing real funds from a Web3 application.

Quick Reference

  • Custodial vs. non-custodial — who actually holds the private key: the platform, or you.
  • FTX (11 Nov 2022 bankruptcy) — ~$10 billion in customer funds lent to Alameda Research; SBF convicted on all 7 counts (2 Nov 2023), sentenced to 25 years (28 Mar 2024).
  • Howey Test (1946) — investment of money, common enterprise, expectation of profit, from others' efforts — the US's own case-by-case security test.
  • MiCA (EU, adopted 20 Apr 2023, applied in full by 30 Dec 2024) — a dedicated regulatory framework for crypto-assets and service providers.
  • Rug pulls and approve() phishing — real Web3 risks that require no smart contract bug at all.