Penny for your thoughts?

A lot of ink has been spilled about Elon Musk's rapidly developing brain implant startup Neuralink, owing in no small part to the kind of hype the billionaire's able to drum up around his companies.

But Neuralink, like its founder, has a little brother: Mark Zuckerberg's Meta — formerly Facebook — which is starting to look like the tortoise to Musk's hare.

Since 2016, Meta's brain-interface technology has developed steadily along two paths: one, a module allowing subjects to "hear with their skin," and the other a "silent speech system" meant to help users communicate with just their brains.

A wearable headband building on the latter venture was scuttled a few years ago, though Meta-backed academic research into brainwave communication quietly continued while the Neuralink circus raged.

Now, Meta says it's taken a huge step forward. For the first time, the company's researchers say they've been able to train an AI model to decode brain waves with up to 80 percent accuracy in laboratory settings.

Basically, the company was able to record the tiny magnetic fields generated by natural electrical currents in the brain so that human subjects could "type" with their minds. Those sentences were then cross referenced with an AI model's readout of the magnetic fields, which were pretty faithful to the original, give or take a few typos.

The mindreading breakthrough is cool in the abstract. But as Vox Future Perfect Fellow Celia Ford points out, we'll soon need to grapple with what it means for Meta to be the tech's gatekeeper — and for a notoriously invasive tech corporation to have access, at least in theory, to the data tucked away in our brains.

Unfortunately, there are plenty of reasons to worry. As a data broker, Meta has been a historically bad actor.

In 2018, a massive New York Times investigation revealed that Meta, then Facebook, was trading user data with third-party companies, mainly other big tech firms. Though it was caught sharing users' full names and contact info, Meta defended its actions, saying the data transfer allowed people to have "more social experiences."

Six years later, it was revealed that Meta had been waging a corporate espionage campaign to decrypt user traffic from Snapchat, YouTube, and Amazon. That project ran from 2016 to 2019, and the company vehemently denied it until federal court records revealed that Zuckerberg himself launched the effort.

Then there was the infamous NSA scandal in 2013, where it was revealed that Meta, along with the likes of Google, Microsoft, and Yahoo, were actively involved in the government collection of private user data. That scandal paled in comparison to the Cambridge Analytica debacle, when data harvested and held by Meta was accessed by third-party actors and sold to foreign agents.

Meta's AI practices are likewise mired in controversy. Though at one time the company prohibited its AI model's use in military applications, it walked that pledge back in late 2024 before partnering to provide AI to a handful of national security administrations and weapons contractors, like the infamous Lockheed Martin.

Its also fed user posts — dating back to 2007 — to its data-hungry  AI model, a practice which most users around the world were unable to opt out of.

So as Meta inches closer to being the mind reader of the future, it's important to remember its track record as the media platform of the past. As it stands, the for-profit tech giant has a lot to prove before we let it loose on our very thoughts — the last truly private domain.

More on Zuckerberg: Zuckerberg Announces Plans to Automate Facebook Coding Jobs With AI


Share This Article