The era of AI hacking is closer than you think

And finally – complexity: AI-powered hacks open the door to more complex strategies than can be devised by the unaided human mind. Sophisticated statistical analyzes of AIs reveal relationships between variables and exploits that even the best strategists and experts never knew existed. That complexity can allow an AIS to deploy mechanisms that compromise multiple levels of the target system. For example, AI designed to maximize a political party’s vote share can pinpoint economic variables, campaign messages, and electoral adjustments that can make the difference between winning and losing an election, extending the revolution that mapping software brought to gerrymandering. All aspects of democracy. And that’s not even getting into the hard-to-detect methods AI suggests for manipulating the stock market, legislative systems, or public opinion.

With the speed, scale, scope and complexity of computers, hacking becomes a problem that we as a society cannot manage.

I remember a scene in the movie. TerminalKyle Reese explains to Sarah Connor that the cyborgs are hunting her: “It’s non-negotiable. cannot be considered. He feels no pity, or remorse, or fear. And it will never stop, ever…” We are not communicating. Literally Cyborg assassins, but as AI becomes our enemy in the world of social hacking, we may find it just as difficult to keep up with its inhuman ability to prey on our vulnerabilities.

Some AI researchers worry about the extent to which powerful AIs can overcome human-imposed constraints and possibly control society. Although this may sound like wild speculation, it is a situation that should at least be considered and prevented.

But today and in the near future, the hackers described in this book are powerful against the rest of us. All AIs, whether on your laptop or online or embedded in a robot, are programmed by other people, usually at your behest, not yours. Although an internet-connected device like Alexa may seem like your trusty friend, never forget that it is designed to sell Amazon products. And just as Amazon’s website encourages you to buy its house brands over competitors’ high-quality items, it won’t always work for you. It destroys your faith in Amazon for its shareholders’ goals.

In the absence of meaningful regulation, there is nothing we can do to prevent AI hacking from unfolding. We must accept its inevitability and build strong governance structures that can respond quickly and effectively by normalizing useful hacks into the system and eliminating malicious or unwittingly harmful ones.

This challenge raises deeper and more serious questions than how AI will change or how institutions can respond to it: What kind of hacking counts as useful? Which ones are affected? And who decides? If you think government should be small enough to drown in a bathtub, you’ll think hacks that reduce the government’s ability to control its citizens are usually good. But you still may not want to replace technology buyers with politics. If you believe in the precautionary principle, you want as many professionals as possible to investigate and prosecute hacks before they become integrated into our social system. And you might want to apply that principle upwards, to the institutions and structures that might be doing those hacks.

The questions continue. Should AI-powered hacks be managed locally or globally? By administrators or by referendum? Or is there a way to let the market or civil society decide? (Current efforts to apply governance models to algorithms are an early indication of how this will play out.) The governance structures we design will give some people and organizations the power to decide the hacks that will shape the future. We must ensure that the power is used wisely.

Taken from The Mind of a Hacker: How the Powerful Bend Community Ruled and How to Bend Back By Bruce Schneier. Copyright © 2023 by Bruce Schneier. Used by permission of the publisher, WW Norton & Company, Inc. All rights reserved.