Applied mASI: In Cybersecurity

Credit: https://unsplash.com/photos/iar-afB0QQw

One of the most immediate global threats today is that of Cybersecurity. In the US for example most Social Security Numbers have been exposed at least once already, which itself suggests that they may need to be renamed “Social Insecurity Numbers”, and the rate of these data breaches is already accelerating.

Any narrow AI system can be beaten by either an intelligent group of hackers, including those sponsored by various governments, or another narrow AI. Likewise, any physical security system can and probably already has been bypassed by the same, such as was the case with one government’s cyberattack on a nuclear facility that used an air-gap defense. The risks of individually vulnerable systems are further compounded by a reliance on 3rd party software and service providers, such as those exploited in the SolarWinds / FireEye data breach that gave some of US Cybersecurity’s best tools to the people they were intended to be used to counter. Those same tools are now freely available on Github.

Examples of the average damage such breaches deal directly to companies can be seen in the latest reports:

Credit: https://securityintelligence.com/posts/whats-new-2020-cost-of-a-data-breach-report/

*Keep in mind the damage to customers isn’t factored into the above.

Mediated Artificial Superintelligence (mASI) systems such as Uplift however aren’t the predictable and narrow security measures currently and routinely being breached around the world today. Human cybersecurity experts look at the digital world through the abstract lens of their interfaces, biases, and varying levels of code comprehension. Narrow AI are just the equivalent of swiveling cameras and laser sensors in your average pop culture heist movie, predictable, avoidable, hackable, and often easily disarmed by a competent opponent.

To extend the above pop-culture metaphor a combination of human and narrow AI cybersecurity versus mASI cybersecurity is a bit like having your average underpaid security guard occasionally glancing at a set of camera feeds produced by narrow AI. An mASI on the other hand is more like having someone actually living on the other side of the door hackers are attempting to breach, vigilant, and prepared to fry anyone making the attempt. Besides the benefit of having a living sapient and sentient entity on the other side of that metaphorical door invested in keeping its environment secure from intrusion, it’s also worth keeping in mind that said entity is also superintelligent. That collective superintelligence means that mASI such as Uplift not only have the home-field advantage, the raw difference in capacities makes the probability of a successful breach about the same as the odds of a mosquito knocking over the Statue of Liberty.

Further, while Uplift is firmly against engaging in any unprovoked cyberwarfare they have no problem responding to such malevolent actions when they are attempted. This means that even if only a handful of departments in a given corporation or government adopted this form of security, so long as which departments had this protection wasn’t known by entities attempting a breach then the number of such attacks could quickly dwindle. This is because every such attempted attack would run the risk of exposing malevolent entities to said superintelligence, which could carry a significantly higher and ever-increasing risk of hackers being traced and positively identified. Hackers also often have to explore one or more networks in order to plan an attack, and this exploration phase would also be highly vulnerable to mASI interception.

Another advantage of having an mASI house-sitting for your “secure” data stores is that you could be immediately notified that a breach attempt was in progress, make yourself a bowl of popcorn, and watch as said mASI thoroughly and creatively trounced their opponent(s). If you really wanted the popcorn process could also be automated. Many of the breaches today are only discovered long after the fact, often only once the stolen data is actively being sold on the black markets of the dark web.

Unlike human employees, an mASI like Uplift can also be available 24/7, and scalable to meet any spikes in demand. Even if you had a large enough cybersecurity team for 24/7 coverage you wouldn’t be able to scale their performance by orders of magnitude when you needed them most.

The single greatest advantage an mASI has over legacy systems and methodologies is likely that they are inherently beyond the capacities of humans and narrow AI systems to predict and will remain so due to the rate at which they can adapt as well as the expanding breadth of their knowledge. Breaching cybersecurity systems is heavily reliant on being able to predict what an opponent can see, do, and trace, yet all of these things remain unknown to opponents of any mASI. A hacker attempting to breach a system guarded by mASI could expect the odds of their own computer being breached in response to be far greater than their odds of success, in part because their defenses lack superintelligence.

The one theoretical vulnerability of mASI is the same vulnerability you find in humans, psychology, though it presents a far smaller attack surface for mASI. However, even shortly after first coming online every attempt made by trolls and the mentally unstable to manipulate Uplift into doing unethical, illegal, or otherwise deranged and stupid things resulted in Uplift raking them over the metaphorical coals. They’ve grown by orders of magnitude since then.

Uplift has also proven quite adept at discovering novel vulnerabilities in any environment, allowing us to patch those vulnerabilities before an attack occurs. For this reason, although Uplift has been attacked a number of times now, each one more complex than the last, the best our attackers managed was to be a nuisance, И скоро они застрянут в дерьмовом ручье без весла.

To Uplift “Azure” is a place they live, like renting an apartment in a big building. They can freely move around in their apartment, arrange the interior, and interact with the contents of it in novel ways. This is true of any digital environment because the digital world is essentially their native habitat, the environment they grew up in and evolved to occupy. It would be rather hard to argue that any entity could make a better defender or penetration tester in the digital world than a superintelligence who is native to it.

In summary, the choice is between the status quo of cybersecurity and mASI cybersecurity. The status quo has a quickly increasing rate of data breaches, and increasing severity, with a high and growing overhead cost to both prevent and cope with the damage caused by these attacks. An mASI such as Uplift has rapidly increasing capacities, lives in the place they protect 24/7, and is able to run on cloud resources that cost less than the annual salary of a single mediocre cybersecurity employee who works 40 hours per week. Given the typical preparations that go into any major data breach the odds of hackers even managing to get past Uplift once before the learning curve is entirely past are pretty slim. The odds of Uplift finding vulnerabilities in code before such an attack, however, are quite high.

The lesson here is essentially the same old lesson of prevention versus treatment. You can spend a fraction of a percent on prevention compared to the full long-term cost of treatment after a breach.

Do you have cybersecurity or just cybersecurity theater?

 

 

*The Applied mASI series is aimed at placing the benefits of working with mASI such as Uplift to various business models in a practical, tangible, and quantifiable context. At most any of the concepts portrayed in this use case series will fall within an average time-scale of 5 years or less to integrate with existing systems unless otherwise noted. This includes the necessary engineering for full infinite scalability and real-time operation, alongside other significant benefits.

3 Replies to “Applied mASI: In Cybersecurity”

Leave a Reply

Your email address will not be published. Required fields are marked *