Google Launches Fight Club for AI Security

Mitch Wagner
7/25/2017

As AI emerges as a leading tool for both security and network attacks, Google is setting artificial intelligences to fight each other to toughen them up.

The competition focuses on the specialized realm of image recognition. Google Brain, Google's machine learning division, is using Kaggle, a platform for data science competitions, to host a competition to prevent attackers from poisoning input data used to train machine learning in image recognition, according to the abstract describing the competition:

"Most existing machine learning classifiers are highly vulnerable to adversarial examples," the abstract explains. "An adversarial example is a sample of input data which has been modified very slightly in a way that is intended to cause a machine learning classifier to misclassify it. In many cases, these modifications can be so subtle that a human observer does not even notice the modification at all, yet the classifier still makes a mistake."

Photo by Lorie Shaull - Own work, CC BY-SA 4.0, Link
Photo by Lorie Shaull - Own work, CC BY-SA 4.0, Link


Keep up with the latest enterprise cloud news and insights. Sign up for the weekly Enterprise Cloud News newsletter.


The competition will proceed along three tracks, two of them to find the best attack system that can trick a machine learning classifier, and a third for the best classifier to defend against attack.

Tricking machine learning systems is nothing new. Spammers evade spam filters by figuring out what patterns the filter's algorithms has been trained to identify. But more recently researchers have shown "that even the smartest algorithms can sometimes be misled in surprising ways. For example, deep-learning algorithms with near-human skill at recognizing objects in images can be fooled by seemingly abstract or random images that exploit the low-level patterns these algorithms look for," according to Technology Review.

"Adversarial machine learning is more difficult to study than conventional machine learning -- it's hard to tell if your attack is strong or if your defense is actually weak," Google Brain researcher Ian Goodfellow tells Technology Review.

The implications of the contest go beyond image recognition.

"Computer security is definitely moving toward machine learning," Goodfellow tells Technology Review. "The bad guys will be using machine learning to automate their attacks, and we will be using machine learning to defend."

"In theory, criminals might also bamboozle voice- and face-recognition systems, or even put up posters to fool the vision systems in self-driving cars, causing them to crash," Technology Review says.

Google announced in March that it plans to acquire Kaggle.

Related posts:

— Mitch Wagner Follow me on Twitter Visit my LinkedIn profile Visit my blog Friend me on Facebook Editor, Enterprise Cloud News

(9)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
kq4ym
kq4ym
8/7/2017 | 10:54:50 AM
Re: AI
One might wonder how much our leaders are followng game theory knowingly or not, and could AI be set up to recognize plots to utilize game theory to mask intrusions when found or at least to prepare for a possible attack.
Ariella
Ariella
8/2/2017 | 12:03:29 PM
Re: AI
@Mhhf1ve In Musk's proposed scenario, it doesn't seem to be that AI is developing its own motive but following its programming's directive with an extreme acceptance of any means to achieve the end desired.  But others have suggested that AI would develop its own indepedent will.
mhhfive
mhhfive
8/2/2017 | 11:47:07 AM
Re: AI
The AI threat to humanity could have no human-like intentions at all.. There's the "paperclip maximizer" story where an AI ends human existence simply as a by-product of being programmed to obtain as many paperclips as possible. We may not understand or know why AI is trying to kill us!

https://www.economist.com/news/special-report/21700762-techies-do-not-believe-artificial-intelligence-will-run-out-control-there-are
Ariella
Ariella
8/2/2017 | 8:58:28 AM
Re: AI
@mhhf1ve I listened to the entire interview, and he definitely said that the war would be started because of an AI goal to maximize revenue -- rather than the other way around. AI could also then feed into systems to help spark the war.
mhhfive
mhhfive
8/1/2017 | 7:04:47 PM
Re: AI
> "a situation in which market manipulation is achieved by starting a war"

Starting a war to manipulate the markets? Hmm. I'd think it'd be the market crash that would cause a war, not the other way around.. but then again.. there was about 9 minutes there between tweets when our president had top US generals worried for a bit.
Ariella
Ariella
7/26/2017 | 5:46:50 PM
Re: AI
Musk takes it a step further in outlining a situation in which market manipulation is achieved by starting a war. It really made me feel like I've seen that movie, though I'm not sure which. Maybe it was just a TV episode. 

I'm really more concerned about the job displacement that would result from the automation enabled by AI. See https://www.wsj.com/articles/robots-are-replacing-workers-where-you-shop-1500456602
Ariella
Ariella
7/26/2017 | 5:46:49 PM
Re: AI
Musk takes it a step further in outlining a situation in which market manipulation is achieved by starting a war. It really made me feel like I've seen that movie, though I'm not sure which. Maybe it was just a TV episode. 

I'm really more concerned about the job displacement that would result from the automation enabled by AI. See https://www.wsj.com/articles/robots-are-replacing-workers-where-you-shop-1500456602
mhhfive
mhhfive
7/26/2017 | 5:34:36 PM
Re: AI
I think Musk and Zuck are arguing over slightly different things. I think Zuck is looking at the short term -- where his own company operates. And Musk is looking a bit further to the future without thinking about the AI developed within his own companies. 

I think I tend to agree with Musk about AI more.. that AI should be pre-emptively regulated before it gets out of hand. Because I could easily see a scenario where AI (unchecked) could be put in control of all kinds of critical infrastructure and financial systems. Perhaps large financial institutions place large bets on the recommendations of AI systems.. without understanding the underlying reasons. That could create a significant market crash that affects the global economy.... And we should probably think about how to avoid that before we're "surprised" by it.

Zuck is a bit dismissive about how widely AI could be implemented, but I think that's the scary part -- that a narrowly trained AI could be put in control of systems that should actualy require more broad intuition and understanding. Human error will put bad AI in charge of something it shouldn't be in control of.... 
Ariella
Ariella
7/26/2017 | 1:11:27 PM
AI
I find it an interesting turn of events to have major figures take public positions about the safety or lack thereof for AI and then have a rebuttal and a "he doesn't know what he's talking about" retort as happened between Musk and Zuckerberg. Do the people here agree more with one than with the other? 
More Blogs from Wagner’s Ring
We're packing our bags for Dallas, for our cleverly named Network Virtualization and SDN Americas conference, but first we sat down to talk about NV, SDN and our favorite travel snack.
Will Apple's new iPhone 11 grow telco profits? Will Apple TV+ compete in a crowded OTT market? And why does everybody have fabulous hair in an Apple TV+ series where civilization has collapsed because everybody in the world is blind?
After years of defending itself against US spying accusations, Huawei claims the US government is spying on it.
The satellite network operator is looking to the Open Networking Automaton Platform (ONAP) to automate connecting its space-based network with terrestrial operators.
VMware's been shopping this summer, buying three cloud and networking startups that will bolster its telco strategy.
Featured Video
Upcoming Live Events
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events