The new Assa Solo was recently introduced in Europe and we believe is the latest Cliq design. We were provided with samples and were able to show a reporter for Wired’s Threat Level how to completely circumvent the electronic credentials in less than thirty seconds, which she easily accomplished. This is the latest and most current example of a failure in security engineering at Assa. The photograph has been edited to prevent visual decoding of the bitting in order to protect the dealer who supplied the lock to us.
We believe there are multiple failures in security engineering by some of the world’s most respected lock manufacturers in conjunction with the deployment of the technology that involve electro-mechanical locks. Potential security vulnerabilities in these locks should cause every security officer and risk assessment team to re-evaluate individual facilities to determine their risk in the event of compromise and their inability to meet certain statutory requirements, such as Sarbanes Oxley or HIPAA.
In response to demonstrations and our disclosures about the bypass of Assa Cliq locks at Defcon 17, the product development manager of Assa in the U.S. told Wired Magazine that “From what I know of the CLIQ technology it can’t be done,” … “And until I’ve seen it done, it can’t be done.”
We believe this statement typifies precisely the problem at Assa Abloy companies: a failure of imagination. It prompted our research and subsequent discovery of multiple vulnerabilities in Cliq, Logic, and NexGen locks. It is this attitude that will continue to allow us to break locks that are represented as the ultimate in security by these companies, and which often provide a false sense of security to the locksmiths and customers that rely upon these products.
Security is ultimately about liability, and such liability is about competent security engineering of locks by their designers. Lock manufacturers are very proficient at making locks work properly. That is what we refer to as mechanical engineering. Unfortunately, the engineering groups for some of the world’s most respected companies may not, in our opinion, have the requisite skills when it comes to security engineering (the design of locks and associated hardware to protect against different methods of bypass). In other words, sometimes they cannot figure out how to open their own locks without the correct key. This is a familiar theme that we have addressed previously, especially with regard to Medeco.
If these companies dispute our contention and claim that they in fact do have the experience in security engineering, then let them explain publicly how their locks can be opened with paper clips, wires, magnets, shock, vibration, and relatively simple tools. Did they design the locks with these attacks in mind, or do they simply not understand them? Either way, we think such lapses in security engineering are inexcusable, demonstrate incompetence, and should subject these companies to liability if they will not voluntarily and retroactively remedy such problems.
DefCon 17 was held in Las Vegas the first week in August. It is the largest security and hacking conference of its kind in the world. While some locksmiths still believe it is simply a gathering of criminals and, as ALOA has labeled its attendees as “persons of questionable character” such descriptions are inaccurate and ill-informed. In fact, the vast majority of participants are professional information technology and security specialists, government agents, law enforcement, and investigative teams. It is the best place to learn about the latest vulnerabilities in cyber systems and security hardware, including locks, and to network with other security professionals.
The world of physical security is rapidly changing and will be dominated by Information Security professionals because of the integration of electro-mechanical and electronic locking systems into an overall security plan, controlled by computer servers and multiple systems. If locksmiths do not become educated in both cyber and physical vulnerabilities, they will soon find themselves relegated to repairing mechanical systems, with an adverse impact on their revenue.
Since 2003, we have presented detailed information each year at DefCon about some aspect of locks and physical security. 2009 was no exception. Tobias Bluzmanis and myself (Matt Fiddler was taken ill just before the conference and could not attend) offered a detailed powerpoint presentation regarding electronic access control systems. More specifically, we examined the Assa Abloy Cliq electro-mechanical locking technology and what we perceive as serious security engineering flaws in many of the locks that are produced by AA companies, including those of Medeco, Mul-T-Lock, Ikon, and Assa.
We also think it is time to set the record straight and speak out against what, in our opinion, we believe constitutes various grades of deficient, negligent, defective, or just plain incompetent security engineering with regard to some of these products, and the legal and security ramifications of such designs. We also want to clear the air about why we have refused to provide any information to any Assa Abloy company regarding our findings.
Background: 2007-2008 Research
During the past year, our team (myself, Tobias Bluzmanis, and Matthew Fiddler) have chosen to concentrate on an intensive research program that begun after our book on Medeco was released in July, 2008. We focused on electro-mechanical locks. That is because Medeco and other AA companies are attempting to move their customers to this newer, more sophisticated, and vastly more expensive technology. So, we thought we would take an in-depth look at this new technology to see just how secure, or insecure it really was.
Mechanical v. Security Engineering
We draw a distinction between mechanical and security engineering. Lock designs must incorporate both mechanical and security engineering. One without the other is dangerous, especially for high security locks and more to the point, electro-mechanical locks.
We have no qualms with the mechanical engineering of any of these locks. They all work, and they work well from an operational standpoint. Mechanical engineers go to school to learn how to make things work. Unfortunately, in my experience, most do not have a clue about security and how to break things, nor about even rudimentary rules of security design. I would urge any design engineer to read Ross Anderson’s book entitled “Security Engineering.” It is the classic text, in its second edition, with regard to systems design, and what can and WILL inevitably go wrong. Its lessons, although primarily focused on the cyber world, are equally applicable to physical hardware design, and especially the integration, which is occurring at an accelerated pace, of hardware and software for security solutions in locking and access control systems.
Our latest research, disclosed at DefCon 17, has yielded surprising results which document and spotlight what we feel are incredible lapses in security engineering. We believe that the design engineers at the Assa Abloy companies who have produced locks that we have evaluated either do not consider the vulnerabilities we identify as significant, or they have no idea what they are or their impact. The legal and ethical question is: to what extent is a company liable to the dealer or consumer for design deficiencies or defects that relate solely to security? This is a complex question, because mechanical and security engineering intersect in the finished product. Is a lock defective if it can be bypassed easily with simple techniques or tools? We believe the answer is yes. Should the manufacturer be liable for such lapses in security engineering? We also believe the answer is yes.
The affected lock manufacturers, which include Medeco, Mul-T-Lock, Assa, Ikon, and possibly some or all of the other Assa Abloy companies, as evidenced by the correspondence from their General Counsel in the United States, seem to believe that virtually all security defects occur because of the continuing “security wars” as I call it, between manufacturers, criminals, hackers, locksmiths and others. So, as the logic continues, the manufacturer will, in time, cure the defect, but has no duty to retroactively fix anything they have already sold. At least, that is my understanding of their position, as repeated in several letters from Medeco, Mul-t-Lock, and Assa Abloy during the past year.
If we can follow their rationale, they believe that security engineering defects occur in the normal course of lock design and development, and that state-of-the-art attacks will be dealt with when they occur, and cannot be anticipated in advance. In the main, I cannot disagree with this logic at all, either from an engineering or legal perspective. What we do disagree with is the notion that a foreseeable security design defect or deficiency that should have been anticipated by those responsible for conceiving of and producing these locks should be treated in the same fashion. Such defects are, in my belief, legally actionable and should subject the manufacturer to liability by dealers and end-users if they do not voluntarily and retroactively remedy the problem at no expense to dealers or consumers.
Even more importantly, such design issues place the locksmith dealer in an untenable position, because they are the ones that are consulting, recommending, selling, and installing these products, and will be the likely defendants in any lawsuits that stem from the security compromise of the locks they sell. Many locksmiths do not have the time, and often the expertise to do their own research into potential security vulnerabilities, especially when their locks are rated by Underwriters Labs, Builders Hardware Manufacturers Association, or other rating organizations in Europe and elsewhere.
When a locksmith sells a cylinder like the Assa Cliq or Medeco Logic for more than six hundred dollars, I think it is fair to expect that such a lock has been thoroughly tested against different security threats. Both the locksmith and consumer have a right to rely upon such an implied representation of suitability for its intended purpose, which is security. Medeco has stated publicly that they rely on internal experts as well as UL and BHMA to determine vulnerabilities and whether their locks are compliant with the standards. Their answer sounds good, but its logic is fatally flawed, and they know it.
UL and BHMA are only allowed to test for certain vulnerabilities, which is precisely the problem with standards. They do not contemplate many methods of bypass, some quite elementary, and so to use them as the ultimate benchmark or authority as to security is not responsible and in our view, can be misleading and reckless. Few if any of the methods that we have disclosed to bypass Medeco, Assa, Ikon, or Mul-T-Lock are addressed in the standards, which is precisely why these companies must have competent security engineers involved in every phase of lock design and testing. Medeco, for example, claims that its locks meet or exceed all applicable high security standards. So what, if the locks can easily be opened by methods not contemplated within the standards?
We were able to simulate the mechanical bitting for Mul-T-Lock Cliq keys. In this photograph, the factory original key that opens the Mul-T-Lock Cliq is shown, together with our simulate key that was cut on a standard interactive blank that should never, according to representations by Mul-T-Lock, open this cylinder. It does, and with no electronic credentials whatsoever, nor audit trail. See quotes from their advertising, below.
Mul-T-Lock, in its latest correspondence of July 30, 2009, stated that their warranty and liability would only extend to locks that are found to be defective “In normal use.” Well, at least that is what I think it said. You can judge for yourself, because in this case, it is unclear whether they will or will not stand behind their products and protect the locksmith and end-user if their locks are found “wanting” with regard to security. Based upon the statements of the General Counsel for Mul-T-Lock in Israel, reprinted below, my question to them and all other companies is quite simple: just what constitutes “normal use” and do you actually believe that you have no liability whatsoever if the lock can be opened with simple techniques, regardless of whether the attack is by insiders or outsiders, and with or without advanced intelligence?
Specifically, do you believe that any bypass techniques that allow your locks to be opened should not be covered by your warranty or that you are not responsible to fix, repair, or replace such deficiencies? Do you not think that the primary purpose of high security locks is to resist attack, as you have stated in prior correspondence to me? Do you not believe, to put it very bluntly, that locks are designed to be screwed with, attacked, tampered with, and that their primary purpose is to resist multiple and different method of attacks?
It would appear that these companies believe that they have no responsibility to retroactively fix anything dealing with security. Yes, they may make changes going forward, and will be glad to sell their customers new locks (and make more money by selling the lock again that should have been designed properly in the first place). But what about all those customers that spent $600 or more for each Cliq or Logic cylinder, and it can be shown to be easily bypassed or set so virtually anyone with the properly bitted (or synthesized) key can open the lock, with or without an audit trail? As Medeco so arrogantly stated in the Slate.com article, “when you buy a Medeco lock, you are not buying a [magazine] subscription.” And what about the locksmiths and dealers that have to answer to their customers? Should they be liable to repair or replace locks with significant security defects, or should they have to tell their customers to throw them away and buy new ones! We don’t think so.
Liability and Security Engineering
The concept of liability, as it applies to locks, is about the requirement that manufacturers disclose to their dealers and end-users any security flaws or potential vulnerabilities that they know, or become aware of. It should follow that a manufacture should immediately notify its dealers and stop selling locks that it knows, or has reason to believe, have significant vulnerabilities that could be exploited by criminals, terrorists, foreign intelligence agencies, or those that would cause harm by exploiting such weaknesses. Similarly, we think that a manufacturer has a duty to understand and find and remedy non state-of-the-art vulnerabilities before they release a product.
We believe that a failure to adhere to this policy constitutes what we call “irresponsible non-disclosure.” It is precisely what occurred, repeatedly, by Medeco and its security engineering with regard to its deadbolt design that we exposed in 2007. They fixed the problem twice, but did they ever tell their dealers to refrain from selling what we demonstrated as defective locks. Nor did they tell their customers that it was a potential threat, as evidenced by several interviews that we conducted and documented with senior customer service technicians at Medeco in 2007. Nor have they ever admitted the problems with bumping, picking, and the ability to compromise their locks through the use of any key within a system that contained the same sidebar code. It is my opinion that they have intentionally misled their dealers and customers with regard to the security vulnerabilities that exist in their locks.
We also believe that a manufacturer should repair or replace locks that they have sold and which contain serious security deficiencies, and they should do so at their expense. Such design deficiencies should not result in the locksmith or end-user being required to purchase new and upgraded locks. Unfortunately, it appears that Assa Abloy, as one of the world’s largest lock conglomerates, and at least some of its companies do not share in this philosophy, as they have so eloquently noted in correspondence and public statements, noted at the end of this article.
Rather, it appears that they believe that lock exploits, such as we have disclosed at DefCon during the past five years, are inherent in the natural progression of lock design and engineering, and that a manufacturer is not liable, either legally or ethically, to fix or replace such defects retroactively. While I believe this is a nice legal theory which has been put forth by the General Counsel for Assa Abloy in the United States, we think it is only partially true, and not responsible. While we concur that new, state-of-the-art attacks that were unknown when a lock was designed and manufactured generally do not subject the manufacturer to liability, I would submit that the result is and should be quite different when the security vulnerability could and should have been discovered by competent engineers that are responsible for security engineering of a product. Example: a design defect that allows a paper clip to bypass the entire audit control feature and credentials security for a Mul-T-Lock or Assa Cliq, or a two-dollar screwdriver to bypass a Medeco deadbolt mortise cylinder.
Electro-Mechanical Lock Design and Cliq Technology
Many lock manufacturers have been touting the advantages of electro-mechanical and electronic access control systems. There is no question that, if properly designed, they can offer the end-user an incredible array of options. The advantages of electronic credentials are obvious, but again, only if the security engineering has been done competently. Otherwise, these locks can create, in my opinion, huge security and liability issues for the manufacturer, dealers, and end-users.
Cliq technology was developed and introduced about 2002. It appears that the system was initially introduced by Ikon, and then adopted by many of the Assa Abloy companies. The core technology consists of a key that contains mechanical bitting and a processor and battery, which communicates with the microprocessor and sidebar-control motor within the lock. When the proper mechanical and electronic credentials are simultaneously presented to the lock, an internal motor is activated, a rotor turns, and a sidebar is allowed to be pushed into the plug. If the key is properly bitted, then the lock can open.
Each lock and key maintains an audit trail of each access or access attempt. This data can be retrieved by a special programming tool and uploaded into a computer for review. Any key in the system can be added or deleted for any lock.
Assa Abloy companies are representing this technology as highly secure, and the “ultimate security solution.” Mul-T-Lock states in its advertising video, which they refused to allow us to show the attendees at DefCon, (claiming it would violate their intellectual property rights, notwithstanding it is on the Internet) “Where security is an issue, compromise is simply not an option.”
Medeco claims in its advertising that its Logic provides “superior protection against unauthorized key copying.”
Mul-T-Lock also says, “In a world increasingly challenged my mounting security threats, the need for comprehensive locking systems has become an essential requirement in virtually every conceivable market sector.” “Each interactive Cliq key contains a unique electronic ID code. It is designated for one individual only, and cannot be duplicated, altered, or corrupted. “
“If the key is not authorized, the mechanical element in the locking system will simply remain locked.”
“Interactive Cliq: unprecedented benefits. The dual patent-protected technologies employed in interactive Cliq represent a truly successful marriage of electrical and mechanical locking systems offering a double layer of impenetrable security.”
“Audit trail control is an absolute necessity if you hope to keep tabs on the efficacy of your locking network…. Interactive Cliq’s control key enables you to easily access precise data from every cylinder in your facility…each key is designated for use by one individual only. If the key is lost, it is simply made obsolete…This enables total control of every key issued to personnel. “
“Interactive Cliq: launching electro-mechanical locking systems to the ultimate level of security.”
We believe such claims are false and misleading and publicly challenge any Assa Abloy company that is making such claims to dispute our findings. We demonstrated that each claim is only partially true, and we believe leaves a false impression with the consumer.
Cliq Technology and Security Engineering
So now we answer our own question: why haven’t we offered to share our research with Medeco, Mul-T-Lock, Ikon, and Assa, with regard to our ability to bypass their Cliq and Logic cylinders by various techniques? The fact is, we offered to do just that. Not once, but many times, but with the following requirements: (1) that the companies would provide us with current lock samples to confirm our research findings, (2) that we would refrain from publishing any information in order that they might confirm and fix the security engineering defects we identify, and (3) we would require that once they confirm the defects, they repair or replace, at their own expense, every lock they have sold to their dealers and end-users that contains the design defects.
And what was the response from Assa Abloy, Medeco, and Mul-T-Lock?
First, they never addressed the issue of supplying samples. Ever. In fact, when I was at the Mul-T-Lock factory in October, 2008, they said they did not have any Cliq locks. End of discussion!
As to agreeing to retroactively fix or replace locks that had security defects, they said that was not going to happen and was unreasonable to require as a precondition for our cooperation.
Finally, they advised that only their internal experts and UL and BHMA were allowed to test their locks. And they said they were not responsible for security defects, because, you know, this is an ongoing issue in lock manufacturing, and, well, nobody really retroactively fixes locks.
This is not quite true. Several companies, both in the U.S. and Europe have done precisely that, and at great cost to themselves. It is the responsible way to do business as a lock manufacturer.
Cliq Technology: What we did and Why it is a Problem
Cliq locks are employed in commercial, government, and residential applications. They are relied upon to protect critical infrastructure and to comply with statutory requirements involving financial institutions, airports, railway, and power generation facilities. If you are a dealer or end-user, you need to understand that we identified several significant vulnerabilities in Cliq and Logic locks which could adversely impact security.
Potential Security Vulnerabilities
OOur research allows us to bypass the security of some Cliq and Logic cylinders to accomplish the following:
Simulate the mechanical portion of the key for Medeco Logic, Assa and Mul-T-Lock Cliq, and Ikon Verso. Plastic keys can be utilized for the Assa Twin and their latest lock, the Solo, which was just released in Europe. Blanks can be modified to simulate Mul-T-Lock keys and allow any number of special blanks to be cut to any bitting;
Utilize a discarded, stolen, or lost key from an Ikon system to compromise other locks in that system, as well as cylinders within a Medeco Logic system, and in similar fashion, to utilize a key from a Medeco Logic system to compromise an Ikon Cliq system;
Change the bitting on a key for an Ikon Cliq or Medeco Logic system to activate the mechanical bitting portion of other systems;
Allow the use of standard Mul-T-Lock non-interactive blanks to open Mul-T-Lock Cliq, because the interactive element is not operable and the mechanical security of the lock is reduced;
Simulate and bypass the electronic credentials for each of the locks listed above;
Trivially bypass the audit trail for each of the locks so that the use of a key is not logged;
Bump open certain of these locks;
Allow an employee to easily bypass a cylinder so that it will accept a key with any credentials. This can occur in certain Mul-T-Lock and Assa versions of Cliq.
We have posted an edited video showing different versions of the Medeco Logic, Assa Cliq, Ikon Cliq, and Mul-T-Lock Cliq being compromised by different attacks. The video does not show the precise techniques to open the locks for obvious reasons. We are sharing that data with affected government agencies and critical customers who are using these locks.
Each of our attacks requires access, at least briefly, to a properly bitted key. However, we have been able to simulate the mechanical bittings for all of these locks.
Admittedly, these attacks all require access to a key with the correct mechanical bitting. However, in many applications, especially government and commercial, a greater threat level exists and is to be expected. Further, the majority of attacks are likely to occur from within an organization, or with the cooperation of an employee, or a person having access.
Lock manufacturers and consumers appear to believe that just because electronic credentials are utilized to open locks, that somehow these locks are inherently more secure. The problem, in our view, is that everyone has forgotten basic security engineering principles: these are still mechanical locks. Although they may employ the additional security layer with the use of electronic credentials, they are still just mechanical locks that rely on moving components to allow them to open.
In our opinion, it is clear that the engineers at Medeco, Mul-T-Lock, Ikon, and Assa have ignored basic security engineering principles, are ignorant of them, or do not understand the potential for compromise of their locks. They clearly have a failure of imagination when it comes to lock design and testing.
While each of these locks are very clever and sophisticated in design, and clearly integrate mechanical and electronic locking systems to a new level, there are, in our opinion, serious deficiencies in each of these technologies that could potentially result in theft, sabotage, vandalism, compromise of critical information, and even loss of life. For that reason, the industry should re-evaluate the efficacy and design of any electronic cylinder and make certain that the essential and critical components of such systems are secure against different methods of attack. While Cliq and other technologies offer the end-user incredible advantages and options, they also offer a prescription for disaster if they are compromised.
We believe these companies should remedy the design issues that we have identified and which will allow their locks to be compromised, and that they should do so retroactively and at their own expense. As a dealer or end-user, we would encourage you to contact the manufacturer and demand to know the following information:
What version of locks do you have installed at your facility, and have they recently been upgraded? We just learned that Mul-T-Lock will be, for at least the fourth time, revising the design of their Cliq. Ask them if the upgrades have been implemented into any new locks that your company is receiving;
What security vulnerabilities have been identified that would allow these locks to be compromised?
What remedies have been taken by the manufacturer to cure the defects?
What does the manufacturer intend to do to insure the security of presently installed cylinders?
How long has the manufacturer been aware of specific methods of bypass of their Cliq or Logic cylinders?
Have the manufacturers notified any dealer, end-user, or government agency with regard to known or potential security vulnerabilities of Cliq or Logic systems?
Has the manufacturer advised their dealers and end-users that in certain keyed-alike systems, the compromise of one key can render the entire facility vulnerable, which would require a replacement of every cylinder in the system?
If you are a dealer or end-user of Cliq or Logic locks, you may contact our office for further information as to the security deficiencies of these locks, possible statutory ramifications for non-compliance, and your legal rights with regard to locks that you have purchased and which have been found to be easily bypassed.
We have tested a limited number of Assa, Mul-T-Lock, Ikon, and Medeco electro-mechanical locks. One or more of these companies may have remedied certain design issues that we have identified in different versions or generations of locks. Each individual customer should determine specific vulnerabilities for the version and brand of lock that they have in service.
QUOTES FROM CORRESPONDENCE THAT WE RECEIVED IN THE PAST YEAR
MUL-T-LOCK GENERAL COUNSEL
“You have misrepresented that Mul-T-Lock’s policy is not to consider replacing or repairing a product which proves to be defective in normal use. This is a gross misrepresentation and not true.”
ASSA ABLOY GENERAL COUNSEL
“All of your accusations and unreasonable demands seem to stem from your mistaken or feigned belief that because a product may under certain limited circumstances be susceptible to a new form of attack. it is somehow rendered “defective.“
® Cliq, Logic, Keymark, and Nexgen are registered trademarks of Assa Abloy companies.