425 PLI/Pat 277
Practising Law Institute
Patents, Copyrights, Trademarks, and Literary Property Course Handbook Series
PLI Order No. G4-3944
November, 1995
Managing the Private Law Library: 1995
*277 BARBARIANS AT THE GATE: INTERNET SECURITY IN THE LAW FIRM/CORPORATE
ENVIRONMENT
Dave James
Copyright (c) 1995, Practising Law Institute
Contents
Introduction
Hackers, crackers, and phreaks
Why the Internet is of particular concern
Vulnerabilities in the law firm/corporate environment
Electronic mail
Mailing lists and usenet
The World-Wide Web
Telnet
FTP
Dial-Up
Internet servers/Web sites
Security strategies
Virus protection
Backups
Encryption
Private-key cryptography
Public-key cryptography
Password protection
Secure(r) channels
Physical security
Information control
The barbarians are at the gate
*280 Introduction
Hackers, crackers, and phreaks
In the early 1980's, when I was beginning my college career and majoring in computer science, I was introduced to a powerful and relatively new operating system called UNIX. On my university's lone UNIX-based minicomputer, files you stored could, by default, be viewed by any other user. It was, of course, possible to prevent others from viewing your files, but this required a conscious effort. Many of us used that machine to connect to the Internet, which at the time was considered by most of my classmates as useful for little more than exchanging e-mail with friends and colleagues at other locations.
UNIX, developed in the late '70s at the legendary Bell Labs in New Jersey, was designed to be both efficient and powerful -- powerful in the sense that users could perform complex tasks with a minimum of commands and processing time. One of its most attractive (and, as it turned out, dangerous) features was the ease with which it could network, i.e., connect machines at different locations. While UNIX was developed largely to control AT&T telephone switches, its power and elegance made it quickly gain popularity in other applications, and, within a few years, it become very popular at universities. The insularity and collegiality both of Bell Labs and of the universities that adopted UNIX fostered a certain trust and cooperation among users. Many programmers began to write top-quality UNIX-based programs, and often used the increasingly popular Internet to distribute them freely. While some security features were built into UNIX, this milieu of trust and cooperation resulted in security being of less than paramount concern. All that changed in 1988.
On the evening of November 2, 1988, a number of UNIX system administrators began to notice that their machines were running more slowly than usual. It quickly became clear that something *281 was seriously wrong -- computers were grinding to a halt, their memory and CPU time being eaten away mysteriously. As system administrators around the country contacted each other, it became clear that this problem wasn't limited to a single machine or network; computers all over the country were succumbing to this mysterious ailment. Programmers worked around the clock to diagnose the problem, and eventually discovered that a program -- later dubbed a "worm" -- was creating and running multiple copies of itself on individual machines and was using the Internet to infect other machines. Many system administrators quickly shut down their machines or disconnected them from the Internet in an attempt to prevent infection or further damage.
By 5:00 AM on November 3, programmers at UC Berkeley had discovered how to slow the worm's spread, and by that evening a way to kill it quickly and effectively had been developed at Purdue and distributed nationwide. The worm was eventually killed, but its death came only after bringing to a halt both the Internet and over 6,000 machines at universities, businesses, and military and government facilities (including RAND, Lawrence Livermore, and the Los Alamos National Laboratories).
This event was not perpetrated by spies or international terrorists, but by a 23-year-old graduate student at Cornell named Robert T. Morris. Morris took advantage of the power of UNIX, certain security holes in commonly used UNIX software, and the trust established between system administrators at different locations not to destroy data or shut down computers, but simply to point out that such security problems existed. He was duly caught, convicted, and sentenced to 3 years' probation, 400 hours of community service, and a fine of $10,500. [FN1]
*282 The proliferation of computers coupled with both an increase of technically proficient users and greater access to the Internet have caused a sharp increase in break-ins and other security problems. The 1990s have seen the escalation of a war between hackers, crackers, and phreaks [FN2] on one side, and computer security experts and law enforcement on the other.
In the early '90s, a hacker group known as the Masters of Deception (MOD) successfully broke into computers owned by Pacific Bell, Bank of America, TRW Information Services, Information America, and a number of other companies. The most infamous hacker of 1995 (and possibly of the entire decade) will likely be 31-year-old Kevin Mitnick who (in a surprisingly long career for a 31 year old) has broken into computers owned the by Department of the Defense and by numerous corporations, infiltrated the computer system of one of the nation's top computer-security experts, and stolen the credit-card numbers of over 20,000 users of Netcom, a large Internet-service provider (including, presumably, our Library's corporate American Express card number).
*283 Mitnick and many MOD members were, eventually, captured by law- enforcement agencies. While high-profile cases such as these have gained notoriety and considerable press, most corporations, universities, and government installations are loathe to draw attention to breaches in their security, and the majority of computer break-ins go unreported or are kept out of the mainstream media.
Computer-security concerns have been compounded (made better or worse, depending on your point of view) by the April 1995 release of SATAN (the Security Analysis Tool for Auditing Networks), written by two well-known security experts. SATAN was created, according to its authors, as a tool to improve network security; they reasoned that by releasing into the general public a program that "knew" many common hacking methods and used them to find security holes, administrators could make their computers more secure. This is, of course, a double-edged sword: SATAN can also teach amateur hackers (those without years of experience and those not allied with hacker groups) robust methods of breaking into systems. [FN3]
What can account for this explosion in computer break-ins? Some common motives are:
1. Stealing money -- Computer systems are used worldwide to transfer money, and these are prime targets for hacking. In a very well-known case that occurred in early 1995, a group of Russian hackers stole over $10 million from Citibank's money-transfer system.
2. Stealing information -- Corporate espionage has been on the rise since the end of the cold war, and information can be of greater value than money in our information economy. If your computer systems hold information valuable to other companies, *284 governments, or individuals, they are potential targets for break-ins. In fact, unless your firm has good intruder- detection systems in place, it's quite possible that you have already experienced undetected break-ins.
3. Personal vendettas -- Disgruntled current or former employees or personal enemies often steal, erase, or change data as a method of revenge.
4. Psychosocial motivations -- Many hackers are motivated not by greed or rancor but by their status among other hackers and the feelings of power and eroticism [FN4] that can come from hacking difficult systems. If your company has a high profile or claims to be hack-proof, then it has probably already been targeted and/or broken into.
Why the Internet is of particular concern
Hackers gain access to computer systems through a variety of channels including the Internet, dial-in lines, and publicly available PCs. The Internet as it exists today is, however, a particularly fertile environment for hackers. Why is this?
The Internet is based on a technology known as packet switching: data sent between machines (e.g., e-mail and Web transactions) *285 is broken up into packets, which are then routed from the sender through various machines until they reach the recipient. Unless users take special steps, packets are sent unencrypted. At many points along the way, it's possible for others to intercept packets containing your data and read them, prevent them from reaching the intended recipient, or alter the contents. [FN5] While it is certainly possible to bug phone or FAX lines or to record private conversations, the tools, knowledge, and access needed to sniff packets are widely available and, as a result, the Internet is a much less secure channel than phone or FAX for exchanging information. [FN6] Ironically, since anyone can connect to the Internet and packets are ephemeral, the Internet is a rather safe environment for the hackers themselves (Mitnick was caught not as a direct result of his Internet use, but by tracing his telephone calls).
Despite the security problems inherent in the Internet, it is quickly becoming an indispensable channel for communications *286 in the legal and corporate environments. In the remainder of this article, I'll first identify the dangers the Internet can present in the typical law firm/corporate environment, and then discuss what steps users of the Internet can take to protect themselves and their firms.
Vulnerabilities in the law firm/corporate environment
Electronic mail
Before the advent of the World-Wide Web and other Internet-based protocols, millions of people worldwide used the Internet to exchange e-mail. E-mail, like other data sent over the Internet, is arranged into packets and routed through various machines until it reaches its destination. As a result, it is just as susceptible to sniffing as is any other communication -- others can potentially read, intercept, or change your message as it is routed through the 'net. Others, both inside and outside your firm, may also be able to read your mail by having access to your computer (do you leave it logged in when you go to lunch or leave the office?) or to another machine (inside the firm or on the Internet) where your mail is stored or forwarded. [FN7]
*287 Another e-mail security issue is spoofing: someone may send a message that appears to be from you, or send a message to you that appears to be from someone it's not. Many software packages (including the very popular Netscape Web browser make spoofing easy. If you have any reason to doubt the origin of an e-mail message, you should verify it with the purported sender by telephone or in person. You can also use information contained in the e-mail's header to help verify the computer system from which it originated. Spoofing and unplanned access can not only cause embarrassment and misunderstandings, but can also result in victims revealing secrets or taking actions based on incorrect information.
It is possible (and common practice on certain mailing lists and newsgroups) to send anonymous messages on the Internet using a remailer. When you send a message to an Internet remailer, the system forwards your message to the recipient without your return address or any other identifying information. For more information more on remailers, look at http://www.stack.urc.tue.nl/ galactus/remailers/ or http://draco.centerline.com: 8080/ franl/crypto/remailers.html.
One of the most effective ways to ensure that electronic communications remain private is to use encryption. While many PC/Network-based e-mail packages have some encryption built in, different packages often won't talk to each other when encryption is involved, and almost all e-mail exchanged on the Internet is sent as plain text. A standard called Privacy-Enhanced Mail (PEM) is becoming more widespread, so sending secure mail over the Internet may become relatively easy in the next year or two. There are many encryption packages available at little or no cost, and, while encrypting your message requires some extra effort, this step simply must be taken when sending confidential information over insecure channels. I discuss encryption in greater detail on page 22 of this article.
*288 E-mail has created a convenient way both to voice one's opinions and to transfer large amounts of text, and both of these present some problems. "Flaming" (i.e., sending negative messages) is a popular pastime; unfortunately, messages that disparage individuals, companies, or products may be considered libelous and may result in litigation irrespective of the sender's intentions. Sending copyrighted materials by e-mail may, of course, violate copyright laws; most individual e-mail messages are also covered by copyright, so forwarding another's message without permission may also violate copyright law. [FN8]
Mailing lists and usenet
In the librarian community, mailing lists have become very popular; in addition to the security concerns posed by e-mail, mailing lists are of particular concern due to the wide distribution of messages sent to them (subscribers to popular lists may number in the thousands). If you reveal a secret or say something foolish in a message addressed to an individual, it's likely that only he or she will see it. If you send the same message to a mailing list or a Usenet newsgroup, hundreds or thousands may see it, and it can easily be copied and sent to other individuals, mailing lists, or newsgroups. A good rule of thumb is to assume that everyone in the world (including bosses, clients, and competitors) will see your message and, as a result, never to send a message you're not willing to have the entire world see.
Mailing lists also pose a particular problem for reference librarians -- a number of lists exist primarily for legal and corporate librarians, and are often used to ask reference questions. It's quite possible that a librarian at your competitor (e.g., opposing counsel) will read your question and can use it to deduce confidential information or strategies. Since I subscribe *289 to a number of mailing lists, I've encountered this situation a few times and have been faced with an ethical dilemma: when I get a message from opposing counsel that seems to deal with a particular case, should I or should I not forward the message to our attorneys handling that case?
The World-Wide Web
In the last couple of years, the Web has become the Internet's hottest topic and biggest consumer of bandwidth. It has also gained the attention of corporations and retailers hoping to capitalize on its popularity to advertise and sell their products. Unfortunately, data transferred on the Web suffers from the same lack of security as does any data sent over the Internet. There has been a great deal of discussion concerning transmission of credit-card numbers on the Web -- as discussed above, it may be relatively easy for others to intercept your credit card number (along, of course, with its expiration date and your name), and go on a shopping spree.
There are two emerging (and to some extent competing) ways to make the Web more secure. Enterprise Integration Technologies (http://www.eit.com) has developed S-HTTP (Secure Hypertext Transfer Protocol), which builds data encryption into Web transactions. Netscape Communications, Inc. (http:// www.netscape.com) has begun implementing a technology called Secure Sockets Layer (SSL), whereby any data being sent via TCP/IP can be encrypted before it leaves your computer. [FN9] While we can expect easy-to-use or transparent Web security in the coming *290 year, neither S-HTTP nor SSL has yet been fully implemented. As a general rule, you currently shouldn't send confidential information over the Web unless you're sure that one of the above protocols is in use.
If your firm has its own Web site or Internet server, it's possible for hackers to gain unauthorized access to this machine, change or delete information on it, and use it to hack into other machines on the Internet. If it's connected to other systems in the firm, they may use it to gain access to your company computers and databases (see the section on Internet servers below).
Telnet
Telnet is often used to access library catalogs and other publicly available databases at remote sites; as such, it does not present a security risk. If, however, you use Telnet to connect to another machine requiring a password, it's possible for sniffers to discover that password and gain access to the remote machine. Let's say, for example, that your firm provides you direct Internet access, but you also have an account with a service provider for personal e-mail. If you Telnet from your firm to your provider, your login and password are being sent over the Internet as plain text and are susceptible to sniffing. If you use one of your provider's dialup numbers instead, your account name and password are most likely being transmitted through local phone company lines and through your service provider's leased lines; both of these channels, while not 100% secure, are considerably harder to hack than the Internet. The other danger with Telnet is that anyone intercepting your data (including, possibly, your competitors) can discover what kind of research you're doing and what information you're retrieving.
*291 FTP
A major concern when using anonymous FTP is viruses; viruses can take control of your system and do anything from occasionally telling you that "Your Computer is Stoned" to destroying data on hard drives. I'll discuss the dangers of viruses and how to protect yourself on page 17 of this article. If you use FTP to connect to a machine requiring a password, or use it to transfer sensitive data, then this information is just as susceptible to being intercepted as it is when you use Telnet.
Dial-Up
While not specifically an Internet issue, many firms allow employees to dial into PCs and networks, and this can pose an even greater security threat than can the Internet. Even if you keep your dialup numbers and passwords private, a determined hacker might be able to get in and steal your data, destroy it, or encrypt it and hold it for ransom (all of these have actually occurred in various corporations across the country). One way to avoid this problem is, of course, not to allow dial-up access. For most firms, however, the benefits of dial-up outweigh the potential security risks. Since "demon dialers" (software that automatically dials every number in a given exchange and logs those that connect to computers) are widely available, simply keeping your dial-up numbers confidential is not an adequate solution. The best defense is good password protection, which I discuss on page 28 of this article.
Internet servers/Web sites
Many law firms are increasing their presence on the Internet: attorneys and librarians want to exchange e-mail with colleagues and clients; they want to use the numerous Internet resources to do research and retrieve documents; many see a domain name and firm Web page as symbols of prestige; marketing departments see opportunities for free advertising space. Setting up a *292 server to provide firm-wide Internet access requires a serious commitment of time and resources, and if your Internet server is connected to individual workstations, file servers, or application servers, security becomes a major issue. It's possible that, even if security steps are taken, a determined hacker can gain access to your firm's computers and data. I can't possibly cover (and don't pretend to know) all the details of setting up a secure server, but I can offer a few pointers:
1. Farm it out -- A number of companies now claim to offer secure Internet access. Their assistance can range from acting as consultants to installing and monitoring most of the equipment at their site and providing a single, secure connection back to your firm. These options can be expensive, but may be worth it in the long run. It takes a specialist to make a system secure: most MIS professionals and librarians don't have enough time to wear the Firm Security Expert hat as well. Even if you were to study security full time, you can rest assured that someone out there has already spent much more time learning how to break in.
2. Rent space on a provider or set up a standalone server -- If your major concern is setting up a Web page (i.e., posting information on the Web about your firm but not necessarily providing firm-wide Internet access), you might consider either renting space for your Web site with an established provider, or setting up a machine that has no connection to the firm's other computers. With both options, hackers may still get into your server, but abusing your Web server is a minor inconvenience compared with the danger of compromised firm-wide computers and databases.
3. Set up a firewall -- If you're faced with the task of setting up a Web server, a firewall is now almost mandatory. A firewall -- one or more computers placed between your server and the Internet -- has the specific task of preventing unauthorized access. Setting up a firewall is a rather specialized task in itself; to get started you might take a look at:
a. William R. Cheswick & Steven M. Bellovin, Firewalls and Internet Security : Repelling the Wily Hacker (Addison-Wesley 1994).
*293 b. The IETF's Site Security Handbook for System and Network Administrators at ftp://ietf.cnri.reston.va.us/internet-drafts/draft-ietf- ssh-handbook-00.txt.
c. Thinking About Firewalls at ftp:// csrc.ncsl.nist.gov/pub/secpubs/fwalls.ps (Postscript version only).
d. Coping With the Threat of Computer Security Incidents : A Primer from Prevention through Recovery at ftp:// csrc.ncsl.nist.gov/pub/secpubs/primer.ps (Postscript version only).
e. The Internet firewall mailing list firewalls-digest@greatcircle.com-- to subscribe, send e-mail to majordomo@greatcircle.com with subscribe firewalls-digest in the body of the message.
4. Don't do it -- Your company may decide that fast, firm-wide access to the Internet isn't worth the risk it poses. Dial-up access is, of course, still a viable (albeit slower) option.
Web sites and Internet servers don't exist in isolation, and Internet security must be considered in the context of a firm's overall security plan. If others (other employees, the public, cleaning crews) have physical access to your computers, or if you allow dial-up access with potential security holes, then someone may easily be able to circumvent any other Internet security you have in place. If you're worried about security in general, you might consider having an outside consultant do a security audit.
Security strategies
It's been said that the only truly secure computer is one that's shut off, locked in a vault, and protected by armed guards. While no system may be 100% secure, there are many steps one can take to ensure a reasonable level of safety and privacy.
*294 Virus protection
If you get a virus on your system, you're largely at the mercy of the programmer who wrote it. He or she may have chosen simply to display a particular message, occasionally to delete a random file, or to delete all the data on your hard disk. Some viruses are time bombs -- nothing much happens at the time of infection, but when a specific event (e.g., a certain date) occurs, the virus is activated and begins its work. The much-publicized Michelangelo virus works this way -- it infects your system, and then on March 6th (Michelangelo's birthday) it reformats your hard drive. Other programs, called Trojan horses, [FN10] masquerade as useful programs, but are something other than what they appear, and do damage when you run them. In recent months, a "new version" of PKZIP (popular file-compression shareware) found its way to some bulletin boards and archives -- this was not, in fact, PKZIP, but a Trojan horse that erased your hard drive and displayed uncomplimentary messages when you ran it.
In general, viruses can be transmitted only in executable files and not in data files. [FN11] For DOS and Windows computers, these usually have extensions of .COM or .EXE, although viruses can be transmitted in other types of files as well (e.g., batch files and *295 .DLL files used by Windows). If a virus has invaded your system, it may infect the boot sector of your hard disk (the boot sector is, in effect, an executable file), floppies you put in the system, any systems into which you put those floppies, and other machines on your network. Some people worry that viruses can be transmitted through e-mail. This is not really a concern unless one attaches an executable file to a message and the recipient runs it. [FN12] There are often reports of viruses that destroy your CPU or other hardware, but there's no way a virus or other program can do significant damage to hardware.
The major vectors for viruses are:
1. Foreign disks -- People often exchange data and programs on floppy disks (e.g., they may bring a game to the office from home, or may take a data file home to work on). If the computer that wrote the file to the disk was infected (actually, if that disk was ever put into an infected computer) then the disk may carry a virus and infect any other computer it is put into.
2. Downloaded files -- The Internet and commercial online services have numerous archives of shareware and freeware. Much of this software does a good job and can save users money; unfortunately, there's also a chance that one of these programs will contain a virus or will be a Trojan horse (as in the PKZIP example above).
3. Computer networks -- If your computer is connected to a file server or to other computers, it's possible for one of those machines to give your machine a virus. Whoever set up and *296 maintains your network should have installed software to protect it against viruses.
How can you tell if you have a virus? Often the first sign is disaster (e.g., a trashed hard disk). I've discovered a number of viruses by noticing inappropriate disk activity (the hard disk light flashing when it shouldn't have). However, since Windows and disk caches have become popular, it's now much more difficult to determine when your system "should" be accessing the hard disk. I've also been tipped off by noticing a change in performance (e.g., programs taking longer than usual to start up), especially under DOS. There are, however, a number of potential reasons for a slowdown, and more than once I've investigated a performance slowdown only to discover that someone has inadvertently turned off the turbo button. Most well-written viruses, however, give little or no indication of their presence until it's too late, and the best way to combat a virus is to avoid getting it in the first place. Some strategies for this are:
1. Use virus-protection software -- Software to scan your system and floppy disks for viruses, and to protect them from becoming infected, is widely available and not too expensive. Microsoft bundles Microsoft Antivirus with MS-DOS 6.x. Other common packages are Norton Antivirus, Central Point Antivirus, and, for Macintosh users, SAM. There are also a number of good shareware programs, including McAfee Scan and Clean, F-PROT, and, for Macintoshes, Disinfectant. [FN13] Since there are currently over 4,000 known viruses and new *297 ones are being created all the time, it's important that you keep your software up to date. For the popular commercial packages (CPAV, MSAV, NAV, and SAM) you can download the latest viral signature files from http://www.symantec.com/virus/virus.html. Most packages allow you to do manual or automatic scans of your hard disk, allow you to scan floppies or individual programs (e.g., one you've just downloaded), will install as a TSR to help prevent infections, and will attempt to remove viruses that are already on your systems or disks.
2. Scan incoming floppies and downloaded programs -- Many firms demand that all incoming floppies be scanned for viruses prior to use. This policy is best carried out by an MIS or computer-support department, and should ideally be done on a single workstation not connected to a network or other computers, and which is dedicated solely to checking disks and downloaded programs for viruses. [FN14]
3. Write protect your floppies -- If you use floppies to transfer or store programs or data files, write protect them whenever you're not writing data to them (to write protect a 3 1/2" disk, slide the small piece of plastic at the back of the disk so the hole is open). If a computer tells you that a disk is write protected or asks you to turn write protection off, and you don't think the system should need to write information to the disk, then this might be a tipoff that you have a virus.
4. Download only from trusted sites, or don't download at all -- All the major online services (e.g., America Online and CompuServe), most large university or government archives, and most large corporate sites scan everything they post for viruses. Of course if someone at the site forgets to check a file or is using outdated software, it's possible for a virus to slip through the cracks, so it's usually a good idea to check software yourself even if you trust the site. If you're downloading *298 from a site you're not familiar with, then you must check everything you download. Another option, of course, is not to download any programs from the Internet or commercial services; as mentioned before, however, shareware and freeware available on these services can save users a considerable amount of money.
Backups
Despite the best security efforts, hackers and viruses sometimes do get in; power outages occur, hard disks die, and users can inadvertently delete files. As a result, firms should establish and follow procedures for backing up important data. Most firms with networks or mainframes already have backup procedures in place (many back up overnight, when usage is low); thus, even if a hacker or virus gets in or a hard disk dies, no more than a day's worth of data will be lost. If you don't do network backups, or if you're storing information on a disk that doesn't get backed up regularly, there are still a number of steps you can take to protect your data.
In recent years, backup hardware (e.g., tape drives) and software have become inexpensive and relatively easy to use. For a few hundred dollars and a bit of effort, you can set up a system whereby all your data is backed up regularly. Your backup hardware should come with a manual to help you establish backup schedules and procedures. If you need to back up data from a number of machines, there are portable drives that you can quickly attach (typically through a parallel or SCSI port) to the machine you want to back up. If you can't afford backup hardware, you can use floppies to back up your data, although this can be tedious and time consuming. It's often easiest to back up the entire contents of your hard disk, but if you're backing up to floppies, or if your tape drive's capacity is lower than your disk drive's and you have all your original program disks, you might consider backing up only your data. In the event of disaster it might, however, be time consuming to reinstall all your programs from the original disks.
*299 You also need to keep your backup tapes or disks physically secure (from both human and non-human threats) and keep copies in a location separate from the machine you've backed up (in cases of fire, flood, etc., if the backup is near the original data source both may be destroyed). You will probably want fast access to the backups, however, so keeping them all off site is impractical; as a result, you should either make duplicate backups or cycle some of the backups offsite while keeping the most current ones onsite. Some systems automatically encrypt the backups, so that, even if they get stolen or duplicated, the thief won't be able to use the data. If you suspect that your data was lost due to a virus you should, of course, check the restored data immediately to ensure that the virus isn't lurking in the backup files ready to re-infect your system.
Encryption
If you need to send confidential information through the Internet or other insecure channels, or if you can't guarantee the physical security of your in- house computers, then the best way to protect your data from unauthorized access is to encrypt it. Many business already encrypt data sent over the Internet, and there are emerging standards for painless encryption on the World-Wide Web and on e-mail sent over the Internet. These standards are not yet widely in use, so if you're not sure an encryption protocol is in effect, it's best to encrypt data yourself. If you need to use cryptography on a regular basis (e.g., if you need to encrypt all outgoing message to a particular client), then you may want to automate the procedure, and both software and hardwarebased systems are available for this purpose. For a good (albeit somewhat biased) overview of cryptography, you can look at RSA's FAQ About Today's Cryptography at http://www. rsa. com/rsalabs/faq/faq_gnrl.html. You can also peruse Berkeley's Cypherpunk home page at ftp:// ftp.csua.berkeley.edu/pub/cypherpunks/Home.html.
You can also obtain software and hardware designed to secure particular files on, or the entire contents of, your hard disk. You *300 should, of course, back up your data and hold on to the backups until you've installed and tested the encryption system. You can also purchase hardware to encrypt both telephone conversations and FAX transmissions.
While there are many encryption systems and products available, the two basic models for data encryption are public key and private key.
Private-key cryptography
With private-key cryptography, the same secret key is used both to encrypt and decrypt a file. This model is appropriate where data won't be sent over insecure channels or where the sender and recipient have a secure opportunity to exchange the key. Examples of this are:
1. If you simply want to encrypt files on your hard disk, and don't need to send them to anyone else.
2. If you're sending the information to someone else, but you have the chance to exchange a key securely, e.g., by phone (assuming your phones aren't tapped) or in person (assuming someone isn't eavesdropping).
There are a number of private-key encryption packages available commercially and as shareware -- it's important, however, that you choose an encryption algorithm that isn't easily broken. The most thoroughly tested algorithm for private-key encryption is likely DES -- DES was invented by IBM in mid-1970s at the request of the National Bureau of Standards and the National Security Agency, and is widely considered to be secure. [FN15] *301 There are a number of software packages available that provide DES encryption. [FN16]
There are some private-key algorithms other than DES that are generally recognized as secure (RC4 and IDEA are two of the most popular). Note that even though a package promises encryption, that doesn't mean it will deliver. One of the most glaring examples of this is WordPerfect -- WordPerfect 5.x claims password protection for files, but this "protection" is easily broken, and there are programs available that can quickly determine a "locked" file's password. [FN17] The producers of encryption software should tell you what encryption method they use, and it should be one that is generally recognized as secure. If a producer says that the encryption method is secret or proprietary, then your data may not be very secure.
If you write your key down or store it on your computer, then anyone finding it can, of course, decrypt your files. One might think that memorizing a single key and using it for all files would be best; unfortunately, this presents a few problems. Since it's often hard to remember a random string of characters, most people tend to use words, names, etc. as passwords. This opens the door to others to guess your key, and using only letters and numbers can significantly reduce the time required for a bruteforce attack (see the section on password protection on page 28 *302 for suggestions on keys and passwords). Further, if you forget the password or are hit by a bus, your data may become inaccessible. If you encrypt many files and use the same password for all of them, then you increase the chances that an attempt to decrypt your files by brute force will be successful. One option is to use a different, randomly generated password for each file, and to store these in a database encrypted with a password that only you (and, possibly, one or two other trusted people) know.
Public-key cryptography
In the mid-1970s, concurrent with the development and acceptance of DES, cryptographers developed the concept of publickey cryptosystems. With public- key cryptography, the sender and receiver need not share a secret key. The receiver generates a pair of keys -- one public, the other private; the receiver can then send the public key over insecure channels or even release it publicly (hence the name "public-key"). Messages encrypted with the receiver's public key can be decrypted only with the corresponding private key. While there are a few different public-key algorithms available, one developed in the late '70s called RSA has become a de facto standard on the Internet and is being built in to a number of products.
An RSA implementation called PGP (Pretty Good Privacy) is available as freeware, although the license agreement prohibits its use for business purposes. You can get more information on PGP at http://www.cis.ohio- state.edu/text/faq/usenet/p g p - faq/top.html; from Simson L. Garfinkel, Pretty Good Privacy: Encryption for Everyone (O'Reilly & Associates, 1995); and from the usenet newsgroup alt.security.pgp. You can download the software itself (for DOS, Windows, OS/2 Unix, Mac, and other platforms, all of which are mutually compatible) by going to http://web.mit.edu/network/pgp-form.html or the Berkeley Cypherpunk Web site (cited on page 22 above). You can get information on Viacrypt's commercial version of PGP at http:// www.getnet.com/viacrypt. For information on other commercial implementations of RSA (and a free evaluation copy *303 of RSA Secure for Windows), go to RSA Data Security, Inc.'s home page at http://www.rsa.com/.
Most public-key products also offer the ability to create digital signatures. A digital signature is like a fingerprint identifying both the sender and the message. [FN18] Since the sender and recipient don't share a secret key, public-key systems can also help prevent senders from repudiating messages that they've actually sent, and prevent recipients from faking messages and then claiming that someone else sent them. [FN19] Of course with any cryptosystem, messages can be both decrypted and faked if one allows his or her key to be compromised.
With any system of encryption, you need to make sure that the person with whom you're corresponding is who he or she claims to be in the first place (for example, John can send Mary a message saying "I'm Ralph and here's my public key -- send me all your data"). Correspondents can avoid this problem by using a digital registry. Typically, one presents proof of identity to a registry along with his or her e-mail address and public key. Potential correspondents can, then, use the registry to verify this information. RSA Data Security, Inc. has set up such a service *304 and, as of this writing, the US Postal Service has proposed that it be the official registry for keys and e-mail addresses.
One drawback of using public-key systems is that encryption and decryption are typically much slower than are private-key systems (depending on the implementation, DES can be from 100 to 10,000 times as fast as RSA). As a result, many cryptography packages use a combination of DES (or another strong private-key system) and a public-key system in an arrangement known as a digital envelope. With a digital envelope, the message itself is typically encrypted with DES using a randomly selected key. The recipient's public key is then used to encrypt only the DES key; when the recipient receives the message, the software uses his or her private key to decrypt the DES key, and then uses the DES key to decrypt the original message.
One weakness of public-key systems is that it may be relatively easy for attackers to decrypt very short messages. In what is known as the "guessed plaintext attack," if one intercepts a short message and can guess the entire contents of the message, then he or she can use the sender's public key to verify that the guess was correct. [FN20] This problem doesn't exist with private-key *305 encryption systems, since the attacker presumably doesn't have the secret key required to encrypt the test messages.
Note that it is illegal to export products using DES, RSA, and most other "strong" encryption algorithms from the United States without a license (they're classified as munitions under the Export Administration Act of 1979, 50 U.S.C. App. §§ 2401-20, and such export licenses are seldom granted), so you might run into difficulties if you want to send encrypted data abroad. This isn't to say, however, that packages using strong encryption (including DES and RSA) aren't produced and sold outside the US. If you need to correspond securely with someone outside the US, you may be able to have him or her obtain a package abroad that is compatible with your's, or buy two copies of a particular package and ship one to you. There's currently quite a bit of opposition to the export ban (especially since products already exist abroad), so the embargo may be relaxed in coming years.
Password protection
In a typical UNIX system allowing passwords of 5-8 characters in the set [A.. Z,a..z,0..9], there are over 200 trillion possible passwords. Nevertheless, the Computer Emergency Response Team (see page 32) has estimated that 80% of all computer security problems are the result of poorly chosen passwords. Some general rules for selecting passwords (which also apply to choosing cryptographic keys are):
1. Don't use passwords that are common names or that appear in any dictionary (including foreign-language or technical dictionaries). There are many programs and databases available that help hackers guess passwords, [FN21] and if you use a *306 password based on any known word, it will likely be included in one or more of these databases.
2. Don't use passwords that are simple variations of dictionary words (e.g., words spelled backwards, letter substitutions).
3. Don't use passwords that are based on your name or initials or those of friends. (In the movie Wargames (dir. John Badham, MGM/UA, 1983), the character played by Matthew Broderick gained access to a Defense Department computer and almost blew up the world by correctly guessing that the computer's principal designer had used his son's name as a password.)
4. Don't use passwords that people can easily watch you type (e.g., 123456 or qwerty).
5. Don't use automatically generated passwords, since one may discover the algorithm your program uses to generate them.
6. Don't write down your password or store it on a disk. (Our friend from Wargames was also able to break into his school's computer system and "improve" his GPA by knowing where the school secretary kept her passwords written down.)
7. Change passwords often. If you're using a UNIX dialup account for Internet access, you can usually do this with the passwd command.
8. Use passwords that are a mix of letters and numbers, and of different cases.
9. Select a password that means something to you alone and that you can remember; for example, the initial letters of a favorite phrase.
If your firm allows dial-in access to its network, there are a number of tools (some of which are known as "tokens") to help keep hackers out. S/Key [FN22] (developed by Bellcore) and other *307 commercial software and hardware use either a one-time password system (in which the password for each login is either taken from a list or based on a combination of a secret key and the password used for the last login), or a challenge and response system. Even if someone monitors the login (e.g., by sniffing) and obtains the password, that particular password will be invalid for future sessions. If you want to secure individual workstations, there are a number of software packages to do so, and many systems have password security built into the CMOS. Unfortunately, putting a password in the CMOS may also lock out those with legitimate access to your machine, e.g., your MIS department (if they need to perform maintenance or upgrades) and even you (if you forget your password). As a result, if you want to password-protect a firm machine, make sure to check with your MIS department before doing so. It may also a be good idea to disable floppy booting on workstations and file servers or to remove floppy drives altogether, since booting from a floppy may let intruders bypass certain security measures. This also may create headaches for you MIS department, so be sure to check with them first.
Secure(r) channels
For some transmission of data, it just doesn't make sense to use the Internet or any other insecure channel. While it's certainly possible to hack into phone lines or leased lines, it's more difficult and occurs less frequently than on the Internet. Commercial e-mail providers (e.g., MCI Mail) are usually more secure than the Internet, provided that both the sender and recipient use the same service. You can also set up a system where the sender's computer dials the receiver's computer and sends data over normal phone lines. If you need to be even more secure, or if you regularly send large amounts of data, you might want to consider getting a leased line.
*308 Physical security
As mentioned before, an Internet security plan must be considered in the context of overall firm and computer security. The strongest firewall in the world won't hold up against someone who simply has physical access to your computers. Do you have publicly available terminals? Do you put outdated password lists in the trash? Who might have access to your workstations and file servers? Current employees? Former employees? Cleaning crews?
There are many products available to help secure computers physically, and most systems come with a key switch that prevents the keyboard from operating (if you look at the back of your PC, there's a good chance that you'll see a key or two dangling). Screws and glue can be removed, and switches can be shorted through; such security measures, then, are roughly equivalent to locking your car door -- someone determined to break in will probably be able to. These types of systems can, however, thwart the "casual" thief or hacker, and are a reasonable first line of defense.
Information control
One way to ensure the security of your information is simply not to give it out. If someone asks for your credit-card number, social-security number, or other confidential information and you have doubts that the information will be transmitted securely and used honestly, just say no. If our Library hadn't provided Netcom with our credit-card number (they charge $2 more per month to send a bill), there would have been no opportunity for Mitnick to get it. Collecting and selling personal information is *309 big business, and much of it is released by individuals voluntarily. [FN23]
In the corporate setting, information control usually involves giving the right to access or change data only to those who require it to do their jobs. As mentioned before, Internet security must be part of a firm's overall security plan. You might, for example, take the necessary steps to have a client send you a sensitive document in a secure fashion. These steps may be worthless, however, if you print the document and leave it where others can get to it or throw it in the trash without shredding it.
The barbarians are at the gate
What do you do when you detect an actual or attempted break-in? What steps do you take if you suspect your system is infected with a virus?
If the crisis involves a break-in, you should contact CERT -- the Computer Emergency Response Team. CERT was set up in 1988 at Carnegie-Mellon University as a result of 1988 Internet worm. They publish advisories, fix security holes, and keep statistics on computer break-ins. You can get their faq at ftp:// cert.org/pub/cert_faq or contact them at cert@cert.org. You can also call them at (412) 268-7090 weekdays 7:30-6:00 ET for normal business or 24 hours a day for emergencies.
You should also report break-ins to law enforcement authorities Most unauthorized access and interception electronic communications *310 have either been specifically outlawed or considered by courts to be covered by previously existing laws. For a good discussion of laws pertaining to the Internet and data transfer, see Richard D. Marks, "Security, Privacy, and Free Expression in the New World of Broadband Networks," Houston Law Review 32 (1995): 501.
The best security measures are prevention and to have a response plan in place. This plan should ideally be set up by your MIS department and/or a consultant, and should be linked both to a firm security policy and to an Internet use policy. [FN24] If you don't have a formal plan in place, you should discuss with your MIS department, network administrator, or service provider what to do in the event of a break-in or virus attack.
The Internet is still relatively new to many people and organizations, and presents a number of challenges. It is, nevertheless, quickly becoming an indispensable tool in law firm and corporate environments, and those who don't acknowledge its presence and importance will be deprived of its benefits. The day may come when security is sufficiently embedded in the technology so that individual users need not concern themselves with it. In the meantime, however, computers attached to the Internet and information sent using it are only as secure as you make them.
FN1. Morris, the son of one of the nation's top computer-security experts, tried to design the worm so that, while it wouldn't be easy to kill, numerous copies would not be propagated on individual computers; an error in his mathematical modeling caused the latter to occur. Morris even tried sending an anonymous message telling system administrators how to kill the worm, but since administrators were disconnecting their computers and the Internet was bogged down with copies of the worm, the message never got through. For a brief description of the worm, its effects, and the prosecution of Morris, see United States v. Morris, 928 F.2d 504 (2nd Cir. 1990), cert. denied, 502 U.S. 817 (1991). For a more detailed discussion of the Worm, see Eugene H. Spafford, "Crisis and Aftermath; The Internet Worm," Communications of the ACM June 1989: 678 and Jonathan Littman, "The Shockwave Rider; Background on Robert T. Morris, Jr., Author of the Internet 'Worm,"' PC/Computing June 1990: 142.
FN2. In its original use, "hacker" was simply a slang term for "programmer"; it has now come loosely to mean anyone who abuses others' computer systems. A cracker is one who specializes in breaking into systems. A phreak is one who specializes in hacking telephone-company computers and switches.
FN3. For more information on SATAN, go to ftp://cert.org/pub/cert_ advisories/CA-95:06.satan. For a list of sites where you can obtain SATAN, send mail to majordomo@wzv.win.tue.nl with get satan mirror-sites in the body of the message.
FN4. Eroticism? In college I knew a number of computer science and engineering students, some of whom (my best friend among them) confessed to me certain sexual feelings associated with math, engineering, and programming. In my best friend's case, this passion (which began at a very early age) was a major factor in his decision to become a mechanical engineer (he now works for NASA). In a review of "Breaking the Code," a stage play starring Derek Jacobi about computer and encryption pioneer Alan Turing, the reviewer writes that "When speaking about math, Mr. Jacobi's accelerating voice makes Turing's academic calling into a consuming erotic passion: the statement that 'Godel's theorem is the most beautiful thing I know' arrives like an orgasm at the climax of a lengthy disquisition." Frank Rich, "Breaking the Code," The New York Times November 16 1987: C-4.
FN5. This is generally known as "packet sniffing" -- for more information on how data sent over the Internet can be intercepted, see the CERT advisory on Ongoing Network Monitoring Attacks at ftp://cert.org/pub/cert_advisories/CA- 94:01.ongoing.network.monitoring.attacks. Packet-sniffing programs are widely available on the Internet, and there's even a package that allows sniffing on Novell networks. It's also possible for someone without authorized access to your internal network to intercept data by either tapping into or attaching a magnetic inducer to your network cabling. If you back up your file servers over a network, it's also possible for someone who normally doesn't have access to particular data to intercept that data as it's being backed up.
FN6. For more information on TCP/IP and security, see Security Problems in the TCP/IP Protocol Suite at ftp://csrc.ncsl.nist.gov/pub/secpubs/ipext.ps (Postscript version only). (If you don't have a Postscript printer, you should be able to print out Postscript files using Ghostscript, which is available at http://www.cs.wisc.edu/ ghost/index.html.)
FN7. A 1993 survey by Macworld found that 22% of businesses engaged in surveillance of their own employees' computer files, e-mail, or voice mail, and the majority of those companies didn't tell employees that they were being surveilled (Charles Piller, "Bosses With X-Ray Eyes," Macworld July 1993: 118; an abbreviated version also appears in the August 4, 1993 edition of The Recorder.) While there are both pending legislation and questions to be settled in the courts, current law seems to favor an employer's right to read employees' e-mail. For a discussion of this question, see Laurie Thomas Lee, "Watch Your E-mail! Employee E-mail Monitoring and Privacy Law in the Age of the 'Electronic Sweatshop,"' John Marshall Law Review 28 (1994): 139. Since many employees use company-provided systems to send and receive personal messages, a certain amount of discretion is appropriate.
FN8. For a discussion of e-mail and copyright, see Maureen A. O'Rourke, "Proprietary Rights in Digital Data," Federal Bar News & Journal 41 (August 1994): 511.
FN9. For more information on S-HTTP, see ftp:// ietf.cnri.reston.va.us/internet-drafts/draft-ietf-wts-shttp-00.txt. For more information on SSL, see http://home.netscape.com/newsref/std/SSL.html. In mid- 1995, there were reports that two Berkeley graduate students had "cracked" SSL. While they did discover a security flaw, it only existed in a limited number of machines and Netscape was able to issue a patch very quickly.
FN10. A Trojan horse is usually not a virus. In general usage, however, "virus" is used to refer to any malevolent program intended to invade or damage your system, and I use it in this sense throughout the article.
FN11. In mid-1995, there were stories on the Internet and in the press about a relatively harmless virus that resides in Word for Windows documents. Word for Windows supports an embedded programming language (i.e., can contain programming code that runs when the file is opened or when some other event takes place). Word documents, then (along with a variety of other "data" files) are bridging the gap between "programs" and "data" and are thus becoming potential vectors for virus attacks.
FN12. The relative difficulty of saving and running an executable file attached to an e-mail message varies greatly among systems. For users of shell Internet accounts, the process might involve saving the e-mail or attachment, uudecoding it, downloading it, and then running it. For users of commercial online services, the process would typically consist of pressing the download button and then running the program after it was sent to your computer. For users of LAN e-mail systems, the process might be as simple as double-clicking on an icon.
FN13. On the Internet you can get McAfee Scan and Clean from http:// www.mcafee.com/ and Disinfectant from ftp.acns.nwu.edu in the /pub/disinfectant/ directory. You can get F-PROT 2.19 from ftp:// ftp.coast.net/SimTel/msdos/virus/fp-291.zip or from any other SimTel mirror site (SimTel archives have a great deal of very useful freeware and shareware -- for a list of SimTel mirrors, send e-mail to listserv @SimTel.Coast.NET with get mirrors.info in the body of the message). Since the file name for F-PROT will change with any new release, look at ftp:// ftp.coast.net/SimTel/msdos/virus/00_index.txt to find the latest file name.
FN14. For more information on establishing a firm-wide policy on viruses, see Organizing a Corporate Anti-virus Effort at ftp:// csrc.ncsl.nist.gov/pub/secpubs/fedeli.txt.
FN15. Computing power has increased dramatically since the '70s, and, as computing power increases, so does the likely success of a "brute force" attack, i.e., one in which every possible key (with a good DES program and well-chosen keys, numbering in the quadrillions) is tried. There are now available programs that use Triple DES or DES+CBC, which can add extra security to your data.
FN16. In addition to commercial packages, I've seen freeware and shareware programs on commercial online services, although I haven't used them much and can't attest to their security. There is freeware DES encryption available for UNIX, which some UNIX-shell Internet service providers have made available -- type apropos " des " or contact your provider to see if they have installed this software.
FN17. One of these is WPCRACK, available at ftp:// ftp.dsi.unimi.it/pub/security/crypt/code/wpcrack.tar.gz (a DOS executable, but you'll need a UNIX machine to extract it).
FN18. For example: with a digital-signature system, when John writes a message, his software uses an algorithm to create a "digest" of the message, uses John's private key to encrypt the fingerprint, and then uses Mary's public key to encrypt the entire message. Upon receipt, Mary's software uses her private key to decrypt the message, uses John's public key to decrypt the message digest, and then verifies that the digest matches the contents of the message.
FN19. With DES, John could conceivably write his own message, encrypt it with the DES key that he and Mary share, and then claim that Mary sent it. Similarly, Mary could send a message, then later disown it and claim that John faked it as described above. With a public-key system, this isn't possible since John doesn't know Mary's private key.
FN20. For example: John sends Mary a long, unencrypted message ending with "Should I do x?" and Mary sends back a message saying nothing more that "Yes." An attacker intercepting the messages could, then, create two messages, "Yes" and "No," use Mary's public key to encrypt both, and see which one matches the message Mary actually sent. Even if the attacker has to generate and test many messages to find an exact match, this could be done very quickly (minutes, hours, or days) compared to the time required to crack Mary's private key (typically many, many years). This sort of attack doesn't compromise Mary's private key, but does allow an attacker to verify the contents of a single message. Adding text that you'd normally add to a message (e.g., your name and address) won't help if the attacker can guess that you'd add this particular text. The best way to avoid this problem is by adding a block of randomly generated characters to short messages (although to avoid confusion you should let the recipient know you're doing this).
FN21. A very popular password-cracking program, Crack, is available from ftp.cert.org in the /pub/tools/crack/ directory. Besides breaking into other systems, you can use this program to test the security of your own passwords.
FN22. For more information, on S/key, look at ftp:// thumper.bellcore.com/pub/nmh/docs/skey.txt. Program files are available on the same server in subdirectories of /pub/nmh.
FN23. Federal statutes prohibit, except in narrow circumstances, companies and other entities from requiring you to provide your social-security number; moreover, a number of states have statutes prohibiting businesses from requiring you to provide personal information such as address and phone number.
FN24. For a brief overview of company-wide Internet-use policies, see "Focus: Internet Policies," Workforce Strategies: A Supplement to BNA's Employee Relations Weekly, 33(13) (August 21, 1995): WS-43.