Editing MD5 (section)

===Collision vulnerabilities===
{{Further|Collision attack}}

In 1996, collisions were found in the compression function of MD5, and [[Hans Dobbertin]] wrote in the [[RSA Laboratories]] technical newsletter, "The presented attack does not yet threaten practical applications of MD5, but it comes rather close ... in the future MD5 should no longer be implemented ... where a collision-resistant hash function is required."<ref>{{Cite journal
|url=ftp://ftp.rsasecurity.com/pub/cryptobytes/crypto2n2.pdf
|journal=RSA Laboratories CryptoBytes
|date=Summer 1996
|volume=2
|issue=2
|page=1
|title=The Status of MD5 After a Recent Attack
|last=Dobbertin
|first=Hans
|access-date=10 August 2010
|quote=The presented attack does not yet threaten practical applications of MD5, but it comes rather close. ....{{sic}} in the future MD5 should no longer be implemented...{{sic}} where a collision-resistant hash function is required.
}}{{Dead link|date=February 2020 |bot=InternetArchiveBot |fix-attempted=yes }}</ref>

In 2005, researchers were able to create pairs of [[PostScript]] documents<ref>{{cite web |url=http://www.schneier.com/blog/archives/2005/06/more_md5_collis.html |title=Schneier on Security: More MD5 Collisions |publisher=Schneier.com |access-date=9 August 2010 |archive-date=11 April 2021 |archive-url=https://web.archive.org/web/20210411035935/https://www.schneier.com/blog/archives/2005/06/more_md5_collis.html |url-status=live }}</ref> and [[X.509]] certificates<ref>{{cite web |url=http://www.win.tue.nl/~bdeweger/CollidingCertificates/ |title=Colliding X.509 Certificates |publisher=Win.tue.nl |access-date=9 August 2010 |archive-date=15 May 2017 |archive-url=https://web.archive.org/web/20170515022608/http://www.win.tue.nl/~bdeweger/CollidingCertificates/ |url-status=live }}</ref> with the same hash. Later that year, MD5's designer Ron Rivest wrote that "md5 and sha1 are both clearly broken (in terms of collision-resistance)".<ref>{{cite web |url=http://mail.python.org/pipermail/python-dev/2005-December/058850.html |title=[Python-Dev&#93; hashlib&nbsp;— faster md5/sha, adds sha256/512 support |date=16 December 2005 |publisher=Mail.python.org |access-date=9 August 2010 |archive-date=6 May 2021 |archive-url=https://web.archive.org/web/20210506232819/https://mail.python.org/pipermail/python-dev/2005-December/058850.html |url-status=live }}</ref>

On 30 December 2008, a group of researchers announced at the 25th [[Chaos Communication Congress]] how they had used MD5 collisions to create an intermediate certificate authority certificate that appeared to be legitimate when checked by its MD5 hash.<ref name="sslHarmful" /> The researchers used a [[PS3 cluster]] at the [[École Polytechnique Fédérale de Lausanne|EPFL]] in [[Lausanne]], Switzerland<ref>{{cite magazine|url=http://blog.wired.com/27bstroke6/2008/12/berlin.html|title=Researchers Use PlayStation Cluster to Forge a Web Skeleton Key|date=31 December 2008|magazine=Wired|access-date=31 December 2008|archive-date=21 April 2009|archive-url=https://web.archive.org/web/20090421023048/http://blog.wired.com/27bstroke6/2008/12/berlin.html|url-status=live}}</ref> to change a normal SSL certificate issued by [[RapidSSL]] into a working [[CA certificate]] for that issuer, which could then be used to create other certificates that would appear to be legitimate and issued by RapidSSL. [[Verisign]], the issuers of RapidSSL certificates, said they stopped issuing new certificates using MD5 as their checksum algorithm for RapidSSL once the vulnerability was announced.<ref>{{cite web|url=https://blogs.verisign.com/ssl-blog/2008/12/on_md5_vulnerabilities_and_mit.php|title=This morning's MD5 attack&nbsp;— resolved|last=Callan|first=Tim|date=31 December 2008|publisher=Verisign|access-date=31 December 2008|archive-url=https://web.archive.org/web/20090116180944/http://blogs.verisign.com/ssl-blog/2008/12/on_md5_vulnerabilities_and_mit.php|archive-date=16 January 2009}}</ref> Although Verisign declined to revoke existing certificates signed using MD5, their response was considered adequate by the authors of the exploit ([[Alexander Sotirov]], [[Marc Stevens (Cryptology)|Marc Stevens]], [[Jacob Appelbaum]], [[Arjen Lenstra]], David Molnar, Dag Arne Osvik, and Benne de Weger).<ref name="sslHarmful" /> Bruce Schneier wrote of the attack that "we already knew that MD5 is a broken hash function" and that "no one should be using MD5 anymore".<ref>{{cite web |author=Bruce Schneier |url=http://www.schneier.com/blog/archives/2008/12/forging_ssl_cer.html |title=Forging SSL Certificates |publisher=Schneier on Security |date=31 December 2008 |access-date=10 April 2014 |archive-date=9 November 2020 |archive-url=https://web.archive.org/web/20201109014745/https://www.schneier.com/blog/archives/2008/12/forging_ssl_cer.html |url-status=live }}</ref> The SSL researchers wrote, "Our desired impact is that Certification Authorities will stop using MD5 in issuing new certificates. We also hope that use of MD5 in other applications will be reconsidered as well."<ref name="sslHarmful" />

In 2012, according to [[Microsoft]], the authors of the [[Flame (malware)|Flame]] malware used an MD5 collision to forge a Windows code-signing certificate.<ref name="foo" />

MD5 uses the [[Merkle–Damgård construction]], so if two prefixes with the same hash can be constructed, a common suffix can be added to both to make the collision more likely to be accepted as valid data by the application using it. Furthermore, current collision-finding techniques allow specifying an arbitrary ''prefix'': an attacker can create two colliding files that both begin with the same content. All the attacker needs to generate two colliding files is a template file with a 128-byte block of data, aligned on a 64-byte boundary, that can be changed freely by the collision-finding algorithm. An example MD5 collision, with the two messages differing in 6 bytes, is:

 d131dd02c5e6eec4 693d9a0698aff95c 2fcab5{{Background color|#87CEEB|8}}712467eab 4004583eb8fb7f89
 55ad340609f4b302 83e4888325{{Background color|#87CEEB|7}}1415a 085125e8f7cdc99f d91dbd{{Background color|#87CEEB|f}}280373c5b
 d8823e3156348f5b ae6dacd436c919c6 dd53e2{{Background color|#87CEEB|b}}487da03fd 02396306d248cda0
 e99f33420f577ee8 ce54b67080{{Background color|#87CEEB|a}}80d1e c69821bcb6a88393 96f965{{Background color|#87CEEB|2}}b6ff72a70

 d131dd02c5e6eec4 693d9a0698aff95c 2fcab5{{Background color|#87CEEB|0}}712467eab 4004583eb8fb7f89
 55ad340609f4b302 83e4888325{{Background color|#87CEEB|f}}1415a 085125e8f7cdc99f d91dbd{{Background color|#87CEEB|7}}280373c5b
 d8823e3156348f5b ae6dacd436c919c6 dd53e2{{Background color|#87CEEB|3}}487da03fd 02396306d248cda0
 e99f33420f577ee8 ce54b67080{{Background color|#87CEEB|2}}80d1e c69821bcb6a88393 96f965{{Background color|#87CEEB|a}}b6ff72a70

Both produce the MD5 hash <code>79054025255fb1a26e4bc422aef54eb4</code>.<ref>{{cite web
|url=http://www.rtfm.com/movabletype/archives/2004_08.html#001055
|title=A real MD5 collision
|author=Eric Rescorla
|date=2004-08-17
|archive-url=https://web.archive.org/web/20140815234704/http://www.rtfm.com/movabletype/archives/2004_08.html#001055
|archive-date=2014-08-15
|work=Educated Guesswork (blog)
|access-date=2015-04-13}}</ref>
The difference between the two samples is that the leading bit in each [[nibble]] has been flipped. For example, the 20th byte (offset 0x13) in the top sample, 0x87, is 10000111 in binary. The leading bit in the byte (also the leading bit in the first nibble) is flipped to make 00000111, which is 0x07, as shown in the lower sample.

Later it was also found to be possible to construct collisions between two files with separately chosen prefixes. This technique was used in the creation of the rogue CA certificate in 2008. A new variant of parallelized collision searching using [[Message Passing Interface|MPI]] was proposed by Anton Kuznetsov in 2014, which allowed finding a collision in 11 hours on a computing cluster.<ref>{{cite web | url=http://eprint.iacr.org/2014/871.pdf | title=An algorithm for MD5 single-block collision attack using high performance computing cluster | publisher=IACR | access-date=2014-11-03 | author=Anton A. Kuznetsov | archive-date=4 June 2016 | archive-url=https://web.archive.org/web/20160604093753/https://eprint.iacr.org/2014/871.pdf | url-status=live }}</ref>