MetaMask vs Phantom vs Rabby: Can Your Wallet Stop a Drainer?

Every wallet comparison eventually lists the same things: how many chains are supported, what the swap fee is, how many networks sit one tap away. Those details matter, and we have run them before. But there is a harder question that decides whether any of it was worth it, and it has nothing to do with how many tokens a wallet can display. It is whether the wallet can stop someone from emptying it.

That question got sharper in 2025. Wallet-drainer phishing stole $83.85 million from 106,106 victims over the year, according to Scam Sniffer’s 2025 report. The encouraging part is that both numbers fell hard, down 83% and 68% respectively from 2024, as Cointelegraph noted. The uncomfortable part is what the decline hides: the attacks did not stop, they got more efficient, and they concentrated on the one screen every self-custody user still has to read alone, the transaction-approval prompt.

So this is not another feature roundup. It is a security-in-practice test of the three software wallets most English-speaking readers actually reach for: MetaMask, Phantom, and Rabby. We look at how each one defends the moment of signing, where each one is still blind, and what none of them can do for you. For the ownership, revenue, and telemetry side of the same three-way race, see our companion piece on who owns your wallet in 2026. This one is about who protects it.

The backdrop, as of 10 September 2026, is a nervy market. Bitcoin opened the day near $78,300 and slipped toward $77,900, Ethereum traded around $2,465, and Solana sat near $104, all drifting lower ahead of US inflation data and the following week’s Federal Reserve meeting, per Yahoo Finance. Drainer activity tracks these cycles closely: Scam Sniffer found that the third quarter of 2025, which held Ethereum’s strongest rally, also produced the year’s heaviest phishing losses at $31.04 million. When prices move and wallets get busy, so do the people trying to drain them.

The Only Test That Actually Matters

Almost nobody who lost money in 2025 lost it because an attacker cracked their private key. Modern seed-phrase cryptography is not the weak point, and no realistic amount of compute is going to brute-force a 24-word recovery phrase. The weak point is authorization. A drainer does not steal your key; it persuades you to sign a message or a transaction that hands your assets away, and it does so on a screen that, without help, often shows little more than a wall of hex and a contract address.

The 2025 data shows the shape of this. The single largest drainer loss of the year was $6.5 million, taken in September through a Permit signature, according to Scam Sniffer; the year before, the largest single loss was $55.48 million. Only 11 cases topped $1 million in 2025, against 30 in 2024. The megatheft is rarer, but the mechanism is the same one that hits an ordinary user for a few thousand dollars: a signature the victim did not understand. That is why the real security test of a wallet is not the length of its feature list. It is what the wallet does in the two seconds before you tap approve.

Three levers decide the outcome. First, can the wallet simulate the transaction and show, in plain language, what will actually leave your account. Second, can it recognize a known-malicious site, contract, or signature and warn you loudly. Third, can it make the request legible enough that you can catch what the software missed. MetaMask, Phantom, and Rabby pull those levers in noticeably different ways.

Meet the Three Contenders

MetaMask is the incumbent. Born inside Consensys in 2016 and now the flagship of the independent MetaMask company created by the 2026 Consensys split, it counts more than 30 million monthly active users and over 100 million lifetime downloads across roughly 190 countries. CoinGecko ranks it the third most popular hot wallet, behind Tangem and Trust Wallet. Once EVM-only, it now spans Ethereum and its layer twos plus Solana, Bitcoin, and Tron. Its default security posture leans on Blockaid.

Phantom is the challenger that grew up on Solana. Founded in 2021 and valued at $3 billion after a $150 million Series C in early 2025, it runs somewhere around 15 million to 17 million monthly users and ranks fifth on CoinGecko’s hot-wallet list. It has expanded to Ethereum, Base, and Bitcoin while trimming chains it no longer wants to support. Its security engine is Blowfish, which Phantom acquired outright.

Rabby is the specialist. Built by the DeBank team, it is open-source, EVM-only, and deliberately narrow: it does not chase Solana or Bitcoin, and it has no token. It does not appear on CoinGecko’s mainstream popularity ranking because it is aimed at DeFi power users rather than first-time buyers. What it has instead is a security model that treats the signing screen as the entire point of the product.

How a Drainer Actually Works in 2026

The classic drainer follows a script. A lure arrives, usually a fake airdrop, a token mint, a support ticket, or a cloned version of a real app. The victim connects a wallet, which by itself is harmless. Then comes the ask: a signature request. Crucially, it is often not a transfer at all. It is an approval, a Permit, or a delegation, a message that grants the attacker’s contract the right to move tokens later, at a time of the attacker’s choosing.

Gasless approvals are the workhorse here. Permit and Permit2 signatures let a user authorize spending without paying gas, which is convenient for real apps and perfect for thieves, because the victim signs an off-chain message that does not look like a payment. Scam Sniffer counted three Permit-based cases among 2025’s large thefts, worth $8.72 million, or about 38% of the losses in the million-dollar-plus bracket. The victim did not send funds; they signed a piece of text, and a sweeper bot did the rest.

This is the heart of the blind-signing problem. Ledger’s chief technology officer, Charles Guillemet, has spent years warning that the core danger is approving what you cannot read: a user has no reliable way to tell whether a given signature is claiming a harmless airdrop or handing a malicious contract permission to drain the wallet, per Cointelegraph. Every defense that follows is an attempt to make that distinction visible before you commit.

Transaction Simulation, the Core Defense

The single most useful thing a modern wallet can do is simulate. Before you sign, the wallet runs the transaction against a copy of current chain state and reports what would happen: which tokens leave, which arrive, which approvals get granted, and whether your net balance drops in a way you did not intend. Done well, simulation turns an opaque signature into a sentence a human can judge. All three wallets now offer some version of it, but they arrived from different directions and stop at different points.

Rabby built its whole reputation on simulation. MetaMask reaches the same goal through its Blockaid integration, and Phantom through Blowfish, the security firm it bought in 2024. The differences that matter are how much detail each shows, whether the code can be independently inspected, and how far coverage extends beyond Ethereum mainnet. The table below summarizes where each one stands.

Rabby: Signing Is the Product

Rabby’s pre-sign simulation is the most detailed of the three. Every transaction is run in a forked environment before you approve it, and the wallet shows the expected token inflows and outflows, the approvals being granted, and the impact on your portfolio, alongside a risk assessment from its built-in security engine. If a site tries to switch you to the wrong network, Rabby switches automatically; if an address is not on your whitelist, it flags the fact. The design assumption is that the signing screen is where security is won or lost, so that screen gets the most attention.

Because Rabby is fully open-source, those claims can be checked rather than trusted. The wallet was audited in 2025 by both SlowMist and Least Authority; SlowMist reported only low-severity findings, and Least Authority flagged a small number of issues, including one medium-severity item tied to QR-code syncing. Rabby also ships quality-of-life security tools that reduce standing risk, including batch approval revocation and, on BNB Chain, an MEV guard that helps shield swaps from the kind of transaction-ordering games we covered in our look at how on-chain value is extracted and reclaimed. At roughly 0.25%, its in-wallet swap fee is also the lowest of the three, per 10pmTrader.

The catch is scope. Rabby is EVM-only. It will not help you on Solana, Bitcoin, Tron, or any non-EVM chain, and it is squarely aimed at users who already know what an approval is. For a DeFi-native trader moving across Ethereum layer twos all day, that focus is a feature. For a newcomer holding a bit of SOL, it is simply the wrong tool.

MetaMask: Blockaid on by Default

MetaMask’s advantage is reach, and its risk is the same thing. As the most widely installed wallet, it is also the most-targeted, so its defenses have to work for tens of millions of people who are not security experts. Its answer is Blockaid, a transaction-screening system that is enabled by default in both the extension and the mobile app and runs across Ethereum, Linea, BNB Chain, Polygon, Arbitrum, Optimism, and Avalanche. When a transaction or signature looks fraudulent, MetaMask shows a warning before you approve, and it does so in a privacy-preserving way that does not hand your transaction details to a third party.

In 2026 MetaMask also shipped live address-poisoning detection, a direct response to the 65.4 million poisoning attempts Blockaid says it has flagged since January 2025, per MetaMask’s June 2026 security report. Older accounts should check one thing, though: alerts that ship on by default for new users were opt-in for some legacy installs, so it is worth confirming that security alerts are switched on in settings. MetaMask’s Snaps system also lets third parties extend its functionality, which is powerful but adds surface area a cautious user should weigh.

Where MetaMask historically trailed Rabby was the richness of the pre-sign picture; Blockaid leans toward alerting and scanning, and the plain-language net-balance preview that Rabby made its signature feature took MetaMask longer to match. The gap has narrowed, but MetaMask also charges the highest in-wallet swap fee of the three, around 0.875%. For a mainstream, multichain user who wants sensible defaults and the broadest app compatibility, it remains the obvious pick; for someone who wants to inspect every byte, it is not the most transparent option.

Phantom: Blowfish Inside

Phantom took a different route to the same goal: it bought its security team. In November 2024 it acquired Blowfish, a firm that, before the deal, said it had prevented more than 2.8 million scams, scanned over 1.3 billion transactions, and protected upwards of $18 billion in assets. Blowfish now powers the transaction previews Phantom shows before you sign, warning on suspicious approvals and simulating expected outcomes across the chains Phantom supports.

Phantom’s leadership framed the acquisition as a security land-grab. “By bringing on Blowfish, we now have the best security team in crypto,” chief executive Brandon Millman said at the time. Blowfish co-founder and chief executive Fabio Berger added that “Being directly embedded within a leading wallet will allow us to improve user security in ways that simply weren’t possible as a standalone company,” per crypto.news. For Phantom’s core audience, Solana-first users who value a clean consumer experience, this is a strong, sensible default.

Two caveats. First, Phantom is closed-source, so its key generation and signing cannot be independently inspected the way Rabby’s can; you are trusting the company and its auditors rather than verifying the code yourself. Second, Solana’s threat model is not identical to Ethereum’s. Solana attacks lean on malicious program interactions and token approvals of their own rather than Permit-style off-chain signatures, and Phantom’s previews are tuned for that world. Its in-wallet swap fee sits around 0.85%, in the same range as MetaMask.

The Industry Fix: Clear Signing

Per-wallet defenses only go so far if the underlying signing format stays unreadable. The industry’s structural answer is clear signing, standardized as ERC-7730, which defines a shared way to turn raw transaction data into human-readable descriptions. Ledger proposed the standard in 2024, and on 12 May 2026 it handed stewardship to the Ethereum Foundation through the Foundation’s Trillion Dollar Security initiative, positioning a neutral party to maintain a public registry of contract descriptors that wallets can draw on.

The Foundation’s announcement, written by Hester Bruikman, put the stakes plainly: “Approving a transaction is meant to be the last line of defense when exercising control over what happens to your assets on the blockchain,” and therefore “What You See Is What You Sign (WYSIWYS) must be our goal, and Clear Signing must be the default.” The updated ERC-7730 v2 adds cross-chain support, and the list of contributors, including MetaMask, Trezor, WalletConnect, and Fireblocks, per crypto.news, tells you this is becoming a shared baseline rather than one vendor’s feature.

For our three wallets, clear signing is the rising tide. MetaMask is an active contributor to the registry; Rabby and Phantom are moving toward richer human-readable descriptions as the standard matures. None of it removes the need to read, but it changes what there is to read, from hex to English, and that is exactly the shift the 2025 loss data argues for.

EIP-7702, the New Attack Surface

Every capability is also an attack surface, and 2026’s clearest example is EIP-7702. Shipped with Ethereum’s Pectra upgrade in May 2025, it lets an ordinary externally owned account temporarily behave like a smart contract for a given transaction, which enables useful features like batching and gas sponsorship. It also creates a brand-new way to lose everything: if an attacker tricks you into signing a delegation to their contract, your account can be made to execute their code.

Security researchers have already documented the abuse. Work presented at the USENIX Security Symposium and reported by NewsBTC found that 63% of EIP-7702 authorization transactions in the analyzed sample were linked to attacker-controlled contracts, with automated draining tied to more than $2.3 million in confirmed thefts; more than 97% of mainnet delegations pointed to a small handful of contract families running near-identical sweeper code. Drainer kits now disguise the delegation as a “wallet security upgrade” or an “AI asset assistant,” per The Cryptonomist. Scam Sniffer separately logged two large EIP-7702 cases worth $2.54 million in 2025.

The three wallets meet this differently. MetaMask leans in, using EIP-7702 as the primary upgrade path for its Smart Accounts, which means its screening has to be sharp about delegations. Rabby supports a limited delegated-EOA model and, true to form, tries to make the delegation legible before you sign. Phantom, anchored in Solana’s account model, is largely outside the 7702 blast radius on its home chain. This is one place where MetaMask’s ambition raises the stakes: the more your wallet can do on your behalf, the more it matters that you can read what you are authorizing.

Approvals and Revokes: the Standing Risk

Not every risk arrives in the moment. Token approvals persist. When you grant a decentralized exchange permission to spend your USDC, that permission usually stays live until you revoke it, and many apps request an unlimited allowance for convenience. Months later, if that contract is exploited or was malicious to begin with, the standing approval is a door left open. A meaningful share of drainer losses trace back not to a fresh signature but to an old one that was never cleaned up.

Good hygiene means reviewing and revoking approvals regularly, and here the wallets differ in how much they help. Rabby builds batch revocation in, so you can cut multiple standing allowances in one pass. MetaMask surfaces approvals and supports revocation, often in concert with external tools. Phantom offers approval management within its more consumer-focused frame. Whichever wallet you use, treating approvals as something to prune, not to set and forget, is one of the highest-return habits in self-custody, and it costs nothing but a few minutes.

Address Poisoning: the Attack Simulation Cannot Catch

Some attacks slip past even the best simulation, because there is nothing malicious to simulate. Address poisoning is the clearest case. The attacker sends you a tiny or zero-value transfer from an address engineered to look almost identical to one you use often, matching the first and last characters. Later, when you go to send funds and copy an address from your history, you copy theirs. The transaction is technically valid, the destination is exactly what you told the wallet to use, and the simulation confirms, correctly, that your tokens will go where you pointed them. The flaw is human, not contractual.

Wallets fight this with pattern detection rather than transaction analysis. MetaMask’s live address-poisoning detection, built on Blockaid’s data, is the most direct answer among the three, and it exists precisely because poisoning volume has been enormous. Phantom and Rabby lean on warnings, whitelists, and address-book features. But the durable defense is behavioral: verify the full address, not just the ends, use saved contacts or name services instead of copying from transaction history, and send a small test amount first when moving anything large. No wallet can fully protect a user who authorizes the wrong recipient.

What Your Wallet Cannot Do For You

Self-custody has no undo button. There is no chargeback, no fraud department, and no support line that can reverse a signed transaction. That is the point of holding your own keys, and it is also the cost. In April 2026, US Securities and Exchange Commission staff clarified that software enabling self-hosted-wallet transactions is not acting as a broker, per CoinDesk, which is good for the legality of these tools but underscores the flip side: there is no regulated intermediary sitting between you and a loss.

When money does go, the aftermath is slow and uncertain. The SEC’s enforcement machine pursues fraud aggressively, as we detailed in our report on crypto enforcement in 2026, but enforcement is not restitution, and stolen funds rarely come back to individual victims. Recovery, when it happens at all, is a technical and legal race run by specialist firms, a process we followed in Halborn and crypto’s recovery race. The realistic conclusion is that prevention is almost the entire game.

That is also why a hot wallet, however well defended, is not the end state for serious balances. The standard escalation is to move long-term holdings behind a hardware wallet, and to put shared or high-value funds under a multisignature setup, so that no single signature can move everything. We laid out how to do that sensibly in our guide to multisig best practices. MetaMask, Phantom, and Rabby all integrate with hardware wallets, which is the single most effective upgrade most users can make: it turns blind signing on a phone into a deliberate confirmation on a dedicated device.

• Read the simulation before every approval; if the wallet cannot show what leaves your account, treat that as a red flag.
• Prune token approvals regularly and avoid unlimited allowances where you can.
• Verify the full destination address, not just the first and last characters, and send a small test transfer for large amounts.
• Keep default security alerts switched on, especially on older MetaMask accounts.
• Move long-term holdings to a hardware wallet or a multisignature setup.

Which Wallet for Which User

There is no single safest wallet, only the safest fit for how you actually use crypto. Rabby gives the most detailed and most inspectable signing experience, and it is the strongest choice for an EVM-native DeFi user who wants to see exactly what every transaction will do and pays the lowest swap fee doing it. Its blind spot is everything outside EVM. Phantom is the cleanest, best-defended experience for Solana-first and consumer users, with Blowfish doing serious work behind a friendly interface, at the cost of closed-source opacity. MetaMask is the multichain generalist with sane defaults, the widest app support, and the largest target on its back, which makes its default-on Blockaid alerts and address-poisoning detection genuinely important.

The Bottom Line

The most encouraging fact in this whole comparison is that drainer losses fell more than 80% in a year while crypto usage did not collapse with them. Part of that is a cooler market for stretches of 2025, but part of it is exactly the defenses examined here: simulation that shows what you are about to sign, alerts that fire on known-bad actors, and a slow, industry-wide move from unreadable hex toward clear signing. The wallets got better, and the numbers followed.

Even so, the constant across every attack in the 2025 data is a human approving something they did not understand. MetaMask, Phantom, and Rabby each shrink that gap in their own way, and any of the three, kept updated with default protections on, is dramatically safer than a wallet from two years ago. But the last line of defense is still the person holding the phone. Read the screen, prune your approvals, verify the address, and move serious money to hardware. On a nervy September morning with inflation data looming and prices sliding, the market will do what it does; the part you control is whether the next signature is one you can actually read.

Frequently Asked Questions

Which wallet is safest: MetaMask, Phantom, or Rabby?

There is no universal winner. Rabby offers the most detailed, open-source transaction simulation and is the strongest choice for EVM DeFi users; Phantom pairs Blowfish-powered previews with a polished consumer experience and is best for Solana-first users; MetaMask ships default Blockaid alerts and address-poisoning detection across the most chains, making it the safest all-round default for multichain generalists. The safest wallet is the one that matches how you actually transact.

What is transaction simulation and why does it matter?

Transaction simulation runs a transaction against current chain state before you sign and shows the expected result: which tokens leave your wallet, which arrive, and which approvals you would grant. It matters because most 2025 thefts came from users approving signatures they did not understand. Simulation turns opaque data into a plain-language preview, so you can catch a drain before it happens rather than after.

Can a wallet stop a drainer or phishing scam on its own?

It can stop many, but not all. Simulation and malicious-site alerts catch a large share of attacks, and drainer losses fell 83% in 2025 partly because of these defenses. But address poisoning and social-engineering tricks can produce a valid transaction the wallet cannot flag, because the user authorized it. The wallet is a strong safety net, not a guarantee; your own review remains the final check.

Is Rabby safe, and is it really open-source?

Rabby is open-source, so its code can be independently reviewed, and it was audited in 2025 by both SlowMist and Least Authority, which reported mostly low-severity findings. Its detailed pre-sign simulation is widely regarded as best-in-class for EVM chains. The main limitation is scope: Rabby only supports EVM networks, so it is not suitable for Solana or Bitcoin holdings.

What is EIP-7702 and why do drainers abuse it?

EIP-7702, introduced in Ethereum’s Pectra upgrade in May 2025, lets a normal wallet temporarily act like a smart contract for a transaction. That enables useful features but also a new scam: if you sign a malicious delegation, your account can run code the attacker controls. Researchers found most 7702 delegations they analyzed pointed to attacker-controlled sweeper contracts, often disguised as a wallet security upgrade. Only sign delegations you fully understand.

By Yuki Tanaka, senior wallets and security correspondent, HOGE Wire.