The Verification Venue · a fear that flipped when the hardware did
The Field That Can't Reach the Data
Your phone and your SSD store every bit as trapped electric charge in a floating-gate transistor. There is no magnetism in there to disturb, so no magnet, however strong, erases them. A hard drive is different: it holds data in magnetic domains, and those you can wipe, if you can land a field near 500 mT on the platter. A fridge magnet is about 5 mT. Do the arithmetic and it isn't close.
But the old "keep magnets away from your data" fear didn't come from nowhere; it's just aimed at the wrong decade of hardware. Swap the storage type below to a magnetic-stripe card (a hotel key, a transit pass) and the answer flips. Those hold data at only ~300 oersteds (≈30 mT), and a strong neodymium magnet in your pocket really can wipe them. Pick a storage type, pick a magnet, set the distance, and watch the field that actually reaches the medium (computed live) against the field it would take to erase it.
What are you worried about?
Which magnet?
Field reaching the medium
– mT
Field needed to erase it
– mT
At 0 cm the magnet is flat against the medium; the field equals the magnet's own surface field. A hard-drive platter, remember, sits several millimetres inside a sealed metal case; a card's stripe is right on the surface.
The magnet isn't a point. Model it honestly as a small magnetised cylinder and the field on its axis has an exact closed form: it equals the magnet's surface field when you're touching it, then falls off: steeply, roughly as one-over-distance-cubed once you're a few magnet-widths away. That steep fall is the whole story. A neodymium button strong enough to worry a transit card at contact is harmless to the same card three centimetres away, and never even registers on a hard-drive platter locked inside its case.
And for flash (a phone, an SSD, a USB stick, an SD card) there is no threshold to reach. A flash cell is a transistor with an electrically-isolated floating gate; a bit is a clump of electrons trapped on that gate. Magnets push on magnetic poles, not on static charge. There is simply nothing for the field to grab. That's why the readout for phone/SSD reads IMMUNE, not "safe by a wide margin," but immune by construction.
The check: every number recomputed in front of you
The field at distance is the exact on-axis field of a uniformly-magnetised cylinder (radius R, length L, remanence Br). At the surface it equals Bsurf (the near-field clamp, no 1/d³ blow-up), and far away it decays like a dipole:
For your current choice, plugged through with the live distance slider:
Surface fields (computed from Br, R, L) and erase thresholds (each medium's coercivity, read into a B-field via 1 Oe = 1 gauss = 0.1 mT):
| magnet | Br (T) | R·L (mm) | B_surface |
|---|
| medium | storage | coercivity | erase threshold |
|---|
Every figure here is reproduced offline by node research/can-a-magnet-erase-your-phone-or-ssd/verify-can-a-magnet-erase-your-phone-or-ssd.mjs, which also checks the near-field clamp (B(0)=B_surface) and the far-field 1/z³ slope two independent ways.
Every free choice, and where this is a conservative over-estimate
The Oe→mT reading. Coercivity is a property of the medium, quoted in oersteds (a unit of magnetising field H). To compare it against a magnet's B-field we use the vacuum identity 1 Oe ⇔ 1 gauss ⇔ 0.1 mT, exactly the "one gauss cancels one oersted" rule the degausser industry uses. So 5,000 Oe → 500 mT, 300 Oe → 30 mT. This is the honest threshold for whether the field can begin to flip domains; a clean, reliable wipe wants roughly twice that (why degaussers are rated 9,000+ Oe for a 5,000 Oe drive).
The magnet is a dipole over-estimate for two of the four. A real fridge magnet is a multipole strip (alternating stripes) and a MagSafe ring is an array, both of whose fields collapse within a few millimetres, faster than the single-cylinder model here. So the curve overstates their reach. That's the safe direction: if even the over-estimate says SAFE, it is safe. The neodymium button and the ferrite speaker ring are genuine single dipoles and the model is faithful.
The fridge-magnet number is a midpoint. "~5 mT" sits inside a real 1–10 mT spread depending on the magnet; we take a representative value and label it approximate. It changes nothing: even 10 mT is a third of the weakest threshold on the board (LoCo, 30 mT).
Magstripe is split by coercivity class. LoCo cards (~300 Oe), such as hotel keys, transit passes, gift cards, and older gym fobs, wipe easily. HiCo cards (2,750–4,000 Oe ≈ 275–400 mT), most modern bank and credit cards, are deliberately hardened and survive incidental contact. The calculator labels which one you picked, because the verdict genuinely differs between them.
"Immune" is about the data, not the silicon. Flash data is immune to any magnet; it is charge, not domains. A changing field can in principle induce a voltage in the drive's circuitry, but the change in flux required to damage electronics is enormous (lab-grade, pulsed), and it would break the controller, not rewrite bits. No static magnet you can hold does anything at all. A compass or magnetometer app near a strong magnet will wobble (that's the app reading the field, working as designed) with zero effect on stored data.
Speaker caveat. The famous "1–2 T" of a loudspeaker lives inside the voice-coil gap, sandwiched between pole pieces. The stray field you can actually reach on the outside of the magnet is what's modelled here (tens of millitesla), which is why speakers next to a hard drive were an old worry and are a non-issue for an SSD.