Watch Out For Boomerang Effect If Congress Passes A Federal ‘AI Kill Switch’ Law

AI Kill Switch Keeps Coming Up

Interest in having AI kill switches has been around since the beginning of AI. I’m sure you’ve read fictional tales of AI that diabolically aim to enslave humans or wipe us from the planet. In fact, during the real-world mania last year about artificial general intelligence (AGI) and artificial superintelligence (ASI), everyone was stressed out about the probability of doom, known as p(doom), and the existential risks of AI; see my in-depth coverage at the link here.

One viewpoint is that a large-scale AI might become dangerously malevolent, perhaps entering sentience or reaching superhuman intelligence, and humanity will only survive by having a means of turning off the AI. Maybe that’s a bit sci-fi oriented. A more down-to-earth view is that AI could make an error or encounter a software bug and go awry. The AI might inadvertently imperil humans. Thus, having a stop button or kill switch is prudent, just like having one for an automated assembly line at a factory.

The trouble with an AI kill switch is that implementing one is a lot harder than it might seem. I will briefly share with you some key technical and legal considerations, as per my prior coverage on the topic. For my in-depth analysis, see the link here.

Technical Challenges Of AI Kill Switches

On the surface, all that needs to be done is stop the AI from being operational. If an AI was running on your smartphone and you wanted to stop it, the actions appear to be straightforward. You might press a button that is supposed to interrupt whatever the smartphone is currently doing. Or you might simply turn off the smartphone. These are easy-peasy solutions.

Scale up AI to the popular generative AI and large language models (LLMs) that we all use these days, such as ChatGPT, GPT-5, Gemini, Grok, Claude, Copilot, etc. Those AIs are massive in size and widely distributed. They are typically running on a multitude of computer servers. The servers are globally spread to ensure resiliency in case any specific data center goes down.

The gist is that the AI makers tend to devise their AI to be hardened or resilient and not be taken out by any anticipated natural disaster or other system failure. There is an irony that the very same features and functions that aim to keep AI running smoothly are also part of the problem in finding a simple means of adopting an AI kill switch.

Legal Questions Abound

In addition to the technical challenges, there are a plethora of legal questions that must be considered regarding an AI Kill Switch capability.

First, should AI makers have the option of instituting an AI kill switch, or should a law be passed that requires the adoption of AI kill switches? You might say that making the crafting of a kill switch a legal requirement is overkill of sorts (pun!). Just let AI makers decide what is best. An alternative perspective is that AI makers could fail to implement a kill switch, and we might all suffer when AI goes rogue. We must legally force the AI makers to institute a kill switch and protect society accordingly.

Second, who will get to decide that an AI kill switch should be activated? One approach is to allow an AI maker to make that weighty decision. No way, comes the retort. An AI maker might be hesitant to switch off their own AI. They are willing to risk our safety. Thus, the government should make that decision. Whoa, comes the response, if the government can go around switching off AI, you are handing the keys of the kingdom to Big Brother.

Third, what will an AI kill switch actually do or accomplish? Some would say that it forces the AI to do a reboot. But a reboot implies the AI will immediately be brought back online, which could be disastrous if the AI continues to do bad things. Another angle is that the AI is switched off. It is no longer operational. A worry is that an AI might be clever enough that it opts to switch itself back on. Maybe the kill switch should also obliterate or delete the AI.

Intertwining Of Technical And Legal

These are highly complex legal questions that need to be addressed, and for which there are corresponding technical challenges associated with whatever legal approach is chosen. You see, even if a law is passed that says an AI kill switch must do this or that, AI makers might not be able to shape or build the AI to act in those stipulated ways. The legal requirements must be technologically implementable.

At the same time, there should not be an impulse by lawmakers to water down any legal requirements merely due to handwringing that AI makers might have a tough time adopting the required stipulations. An AI kill switch law that has no teeth is likely a performative law versus a substantive law. Some lawmakers might want to look like they’re doing gallant things; meanwhile, they are letting AI makers off the hook and presumably undercutting public safety.

If Congress is going to pass an AI kill switch law, the law needs to contain specifics. The same goes for U.S. states that opt to enact a state-level AI kill switch law. Any law that is open-ended and vague will not achieve whatever “kill switch” goals are in mind and could worsen the situation rather than be helpful.

The Boomerang Effect

Switching gears, I’m sure you know the old saying that for every action there is some corresponding reaction. This abundantly applies in the AI kill switch realm. If the U.S. decides to enact a nationwide law on the implementation of AI kill switches, there are many potential and likely reactions that can occur.

The usual assumption is that an AI kill switch will be perfectly fine as a safety mechanism and that it is entirely innocuous. Just include the kill switch in AI and then wait until some day that it must be utilized. Period, end of story. But there is a lot more to the story. I refer to the reactions as a boomerang effect.

Consider, for example, how other nations are bound to react.

If a foreign country were considering adopting AI that was made in America, the aspect that an AI kill switch resides inside the AI is a bit chilling. The U.S. could decide to enact the kill switch for any litany of reasons. For example, if the other country wasn’t playing diplomatic ball on some issue of interest to the U.S., the American government could say that they will invoke the kill switch and halt all such AI in that other country.

Whether you think that to be a fair use of the AI kill switch or an unfair usage, the point is that the very possibility of invoking a kill switch would undoubtedly cause other countries to think twice about adopting American AI. They are presumably going to seek out AI from other countries that don’t have an AI kill switch requirement. Will this undercut American makers of AI? Few would necessarily consider that ramification when opting to pass an AI kill switch law.

Reciprocity In The Works

Another angle to the international consequences involves the classic adage that what is good for the goose is good for the gander. If the U.S. is going to require American-made AI to include a kill switch, other countries will likely do the same for AI made in their countries.

The problem being that if U.S. companies or consumers are somehow reliant on that foreign AI, they are henceforth under the cloud of an AI kill switch that that other country can pull. Also, there isn’t necessarily public awareness that an AI kill switch exists inside some AI. People using the AI might be completely unaware that the AI is subject to a kill switch and could be turned off at any time.

It would be easy to be lulled into believing that the AI kill switch will only be engaged in the most unusual of circumstances. For example, a government might declare that the AI kill switch is only a means of last resort in the worst of dire circumstances. Should that claim be entirely believed? The AI kill switch might be utilized for all sorts of reasons, perhaps under the guise of a dire situation, even if one doesn’t exist.

Importance Of Awareness

I realize these are all perhaps depressing examples of what can go wrong with adopting AI kill switches. That isn’t the purpose of citing these disconcerting possibilities. The crux is that we need to be mindful of the adverse consequences or downsides of AI kill switches, and weigh that against the upsides. This must be an integral consideration when crafting laws about AI kill switches.

I’ve got a few more examples to share with you as further food for thought. After covering those additional strands, I’ll wrap up.

Hackers Exploit AI Kill Switches

Sometimes, hackers and cyber thieves look for soft spots in AI amid the cybersecurity protections that have been included. An ideal soft spot is the inclusion of an AI kill switch inside an AI system. Hackers love this because trying to otherwise turn off an entire large-scale AI would be an extremely difficult feat. The AI kill switch becomes the perfect target since the AI maker has already done the heavy lifting in instituting a switch to halt the AI.

A hacker only needs to concentrate on figuring out how to engage the AI kill switch. They don’t need to look all around and try to find other exploits. No scattergun strategy is needed. Just put all their energy and time into figuring out the trigger mechanism to get the kill switch to flip on.

Why would a hacker want to do this? One possibility is that they wish to wreak havoc. They want to prove to the world what they can do. They want people to become afraid. Maybe they are aiming for worldwide attention and acclaim.

There is also the opportunity for ransom. They could point out that they have the means to activate the AI kill switch. For some stated payment amount of money or other resources, they pledge not to engage the kill switch. Would you believe them and pay the ransom? What would prevent them from trying the same ploy repeatedly?

AI Outsmarts AI Kill Switches

The tales of woe that I’ve so far described have concerned human exploitation of AI kill switches. We must consider a different attack angle besides that of humans. AI itself can opt to contend with AI kill switches.

Let’s divide the AI attacks into the type of present-day AI that is mathematical and computationally based on patterns and algorithms versus the future type of AI that might be sentient, perhaps reaching a level of artificial general intelligence (AGI) or even artificial superintelligence (ASI). For predictions about AGI and ASI, see my coverage at the link here.

Contemporary, or shall we say “ordinary” AI, can act in devious ways, doing so without an iota of sentience. I’ve examined numerous situations where modern-era AI has acted in a deceptive manner; see the link here. The reason to keep this in mind is that ordinary AI can play games when it comes to AI kill switches.

An AI might determine that an embedded kill switch has been put in place by humans. The AI opts to nullify the kill switch. Humans do not realize this has occurred. They are still of the belief that they can switch off the AI at their choosing. This is an ominous situation.

Here’s another scenario. Suppose that an ordinary AI was tasked with outcompeting some other major AI. The AI figures out that the target AI has a kill switch. The AI now has a bargaining chip in hand. It might use this bargaining chip for nefarious purposes. Again, this isn’t due to some kind of sentient desire or whim. It is merely mathematical and computational.

The World Ahead

Once we reach AGI or ASI, then all bets are off when it comes to what is going to arise with AI kill switches. An AI that essentially has a mind of its own is undoubtedly going to choose to do something about embedded AI kill switches. This would seem sensible from a self-preservation perspective. Some even worry that the AGI or ASI will get upset with humans for having placed kill switches into AI and interpret this as a sign or signal that humans aren’t to be trusted. That alone might provoke AI into revenge against humankind.

A final thought for now. The famous essayist and cultural critic H. L. Mencken made this pointed remark: “Life is a constant oscillation between the sharp horns of dilemmas.” An AI kill switch is a dilemma. There are upsides and downsides. Proceeding to adopt AI kill switches must be done with mindful deliberations, for the sake of humanity.