Threat Intelligence

10/10/2017
09:15 AM
Connect Directly
Twitter
LinkedIn
Google+
RSS
E-Mail
50%
50%

Artificial Intelligence: Experts Talk Ethical, Security Concerns

Global leaders weigh the benefits and dangers of a future in which AI plays a greater role in business and security strategy.

CYBERSEC EUROPEAN CYBERSECURITY FORUM - Kraków, Poland - The future of artificial intelligence was a hot topic at the third annual CYBERSEC Cybersecurity Forum, where security professionals representing Poland, the Netherlands, Germany, and the United Kingdom discussed the pitfalls and potential of AI, and its role in the enterprise.

Is it too soon to have this discussion? Absolutely not, said Axel Petri, SVP for group security governance at Deutsche Telekom AG. "Now is the time to ask the questions we'll have answers for in ten, twenty years," he added. Cybersecurity supported by AI and machine learning can leverage data to generate more insight and fight fraud.

"You are able to use the workforce you have in a smarter and better way by using AI," he said. "How nice would it be if we could have a junior SOC analyst act as well as the smartest guy in the SOC, of which you currently have very few?"

Andrzej Zybertowicz, research fellow at Nicolaus Copernicus University and social adviser to the President of Poland, explained that while locally used artificial intelligence can increase cybersecurity, the effects are "potentially disastrous" on a global scale. It's time to discuss the broad risks of AI and regulations to avoid them, he said, and others agreed.

"The problem is, there are so many risks," said Noel Sharkey, professor of artificial intelligence and robotics at the University of Sheffield, who believes there are both opportunities and threats in the field. "There's not just one thing."

Sharkey presented an example in the medical field, where AI could help doctors research diseases. This is a good thing, he said, but what happens when the machine is right long enough and the doctor stops questioning it? Should a doctor automatically agree with a machine? What are the implications if they do, and the machine someday gets it wrong?

"What's core is making sure there's clear accountability, and being concerned with the types of controls we seek in AI," Sharkey continued. There is a need for deep learning, and deep reinforcement learning, as we seek AI applications in child care, elder care, transport, and agriculture. "Future-proofing" AI should consider its implications for human rights.

"Artificial intelligence transforms everything around us; every industry, our health, our education," explained Aleksandra Przegalinska-Skierkowska, assistant professor at Kozminski University and research fellow for Collective Intelligence at MIT. "Especially if we want autonomous vehicles or virtual agents, we need a code of conduct for them."

We are at a point when people have begun to reflect on issues related to machine ethics and morality, she added. Building a structure for ethical AI systems should be a collaborative effort, especially as more businesses generate connected products.

"From the perspective of a company selling digital services, we should put one very important thing at the center of our attention -- this is a customer using the AI," said Petri. "What we need is the trust of users in every technical system. If we don't have they trust, we don't have users."

The discussion of regulatory measures soon turned to threat actors who will break them. Almost every technology is dual-usage and can be weaponized, Zybertowicz pointed out. "We are here talking about rules, but we are dealing with a group of bad actors who don't play by the rules," said Petri. "The bad guys are innovating faster than the good guys."

Related Content:

Join Dark Reading LIVE for two days of practical cyber defense discussions. Learn from the industry’s most knowledgeable IT security experts. Check out the INsecurity agenda here.

Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
Why CISOs Need a Security Reality Check
Joel Fulton, Chief Information Security Officer for Splunk,  6/13/2018
Cisco Talos Summit: Network Defenders Not Serious Enough About Attacks
Curtis Franklin Jr., Senior Editor at Dark Reading,  6/13/2018
Meet 'Bro': The Best-Kept Secret of Network Security
Greg Bell, CEO, Corelight,  6/14/2018
Register for Dark Reading Newsletters
White Papers
Video
Cartoon
Current Issue
Flash Poll
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2018-12294
PUBLISHED: 2018-06-19
WebCore/platform/graphics/texmap/TextureMapperLayer.cpp in WebKit, as used in WebKitGTK+ prior to version 2.20.2, is vulnerable to a use after free for a WebCore::TextureMapperLayer object.
CVE-2018-12519
PUBLISHED: 2018-06-19
An issue was discovered in ShopNx through 2017-11-17. The vulnerability allows a remote attacker to upload any malicious file to a Node.js application. An attacker can upload a malicious HTML file that contains a JavaScript payload to steal a user's credentials.
CVE-2018-12588
PUBLISHED: 2018-06-19
Cross-site scripting (XSS) vulnerability in templates/frontend/pages/searchResults.tpl in Public Knowledge Project (PKP) Open Monograph Press (OMP) v1.2.0 through 3.1.1-1 before 3.1.1-2 allows remote attackers to inject arbitrary web script or HTML via the catalog.noTitlesSearch parameter (aka the S...
CVE-2018-10811
PUBLISHED: 2018-06-19
strongSwan 5.6.0 and older allows Remote Denial of Service because of Missing Initialization of a Variable.
CVE-2018-10945
PUBLISHED: 2018-06-19
The mg_handle_cgi function in mongoose.c in Mongoose 6.11 allows remote attackers to cause a denial of service (heap-based buffer over-read and application crash, or NULL pointer dereference) via an HTTP request, related to the mbuf_insert function.