Cryptography : how to protect critical systems in the quantum era

The advent of quan­tum com­pu­ting repre­sents a major tech­no­lo­gi­cal revo­lu­tion, but also an unpre­ce­den­ted threat to digi­tal secu­ri­ty. First of all, quan­tum com­pu­ting holds great poten­tial in terms of data sto­rage and com­pu­ting power, as well as mul­tiple appli­ca­tions in the fields of simu­la­tion and opti­mi­sa­tion, with revo­lu­tio­na­ry pros­pects in the che­mi­cal, phar­ma­ceu­ti­cal and digi­tal indus­tries for example¹. It the­re­fore repre­sents a tech­no­lo­gi­cal break­through that could bring many bene­fits to a socie­ty cur­rent­ly facing major eco­no­mic and socie­tal chal­lenges².

Asym­me­tric encryp­tion algo­rithms, com­mon­ly used to secure elec­tro­nic com­mu­ni­ca­tions, are par­ti­cu­lar­ly vul­ne­rable. While conven­tio­nal com­pu­ters take thou­sands of years to solve these pro­blems, a quan­tum com­pu­ter could do so in just a few hours, as illus­tra­ted by Shor’s algo­rithm des­cri­bed in 1994³, which is capable of fac­to­ring large num­bers expo­nen­tial­ly fas­ter than conven­tio­nal methods. In fact, the fac­to­ri­sa­tion pro­blem posed by cer­tain encryp­tion algo­rithms (in par­ti­cu­lar RSA) can be trans­for­med into a pro­blem of fin­ding a per­iod in a func­tion ; sol­ving this second pro­blem is acce­le­ra­ted by the super­po­si­tion of quan­tum states⁴. Simi­lar­ly, Gro­ver’s algo­rithm⁵ threa­tens sym­me­tric encryp­tion mecha­nisms (AES) and hash func­tions (SHA).

The deve­lop­ment of machines capable of exploi­ting this vul­ne­ra­bi­li­ty could have disas­trous conse­quences for govern­ments. These could include acts of indus­trial espio­nage, sabo­tage of cri­ti­cal sys­tems, iden­ti­ty theft and even mani­pu­la­tion of elec­tions, with signi­fi­cant reper­cus­sions for natio­nal secu­ri­ty and social sta­bi­li­ty. Howe­ver, in 2025, quan­tum com­pu­ters are still at the expe­ri­men­tal stage. Only orga­ni­sa­tions with consi­de­rable resources and advan­ced tech­no­lo­gi­cal skills are capable of imple­men­ting these actions⁶. To put it sim­ply, govern­ments need to pro­tect them­selves against powers capable of deve­lo­ping “nuclear weapons”.

To deal with this threat, it is impe­ra­tive that govern­ments anti­ci­pate and imple­ment a stra­te­gy for the tran­si­tion to post-quan­tum cryp­to­gra­phy. This approach involves deve­lo­ping encryp­tion algo­rithms that are resis­tant to quan­tum attacks, the­re­by gua­ran­teeing the secu­ri­ty of com­mu­ni­ca­tions in the era of quan­tum com­pu­ting. These algo­rithms must not be based on dis­crete fac­to­ri­sa­tion or loga­rithm pro­blems, which are vul­ne­rable to quan­tum technology.

The US Natio­nal Ins­ti­tute of Stan­dards and Tech­no­lo­gy (NIST) has selec­ted three of the most secure and effi­cient post-quan­tum algo­rithms fol­lo­wing a rigo­rous selec­tion pro­cess⁷ :

• ML-KEM (ori­gi­nal­ly known as Crys­tals-Kyber), desi­gned to secure access to sites via a public channel.
• ML-DSA (for­mer­ly Crys­tals-Dili­thium), which gene­rates elec­tro­nic signa­ture keys for secure docu­ment exchanges and communications.
• SLH-DSA (for­mer­ly Sphincs+), which gene­rates smal­ler public elec­tro­nic signa­ture keys.

The ML-KEM and ML-DSA algo­rithms are based on the dif­fi­cul­ty of fin­ding short vec­tors in a struc­tu­red Eucli­dean net­work. Fur­ther­more, as explai­ned in an ANSSI (Agence natio­nale de la sécu­ri­té des sys­tèmes d’in­for­ma­tion) ana­ly­sis note⁸, there is a pos­si­bi­li­ty that a weak­ness will be dis­co­ve­red, enabling a rapid reso­lu­tion of the cryp­to­gra­phic pro­blem posed. The SLH-DSA algo­rithm is based on the secu­ri­ty of hash functions.

In addi­tion, ANSSI recom­mends the use of hybri­di­sa­tion, which consists of com­bi­ning post-quan­tum asym­me­tric algo­rithms, still under deve­lop­ment, with well-esta­bli­shed and pro­ven tra­di­tio­nal asym­me­tric encryp­tion methods. This com­bi­na­tion offers double pro­tec­tion until post-quan­tum algo­rithms reach a suf­fi­cient level of matu­ri­ty to gua­ran­tee long-term secu­ri­ty on their own.

The tran­si­tion to these new algo­rithms requires a great deal of work and will the­re­fore be gra­dual. It will take seve­ral years to replace all exis­ting sys­tems with secure solu­tions. To achieve this, a signi­fi­cant invest­ment in the cryp­to­gra­phic skills of the staff res­pon­sible for pro­tec­ting sys­tems, data and digi­tal exchanges must be made now, over a mul­ti-year per­iod. Indeed, a later invest­ment is like­ly to gene­rate very high adap­ta­tion efforts in a par­ti­cu­lar­ly constrai­ned time­frame, with high risks vis-à-vis exter­nal entities.

To gua­ran­tee the secu­ri­ty of state ser­vices, a tho­rough assess­ment of the risks asso­cia­ted with quan­tum tech­no­lo­gies must be car­ried out. This involves not only unders­tan­ding the capa­bi­li­ties of quan­tum com­pu­ters but also asses­sing the poten­tial impact on exis­ting secu­ri­ty sys­tems. In par­ti­cu­lar, it is essen­tial to iden­ti­fy the most sen­si­tive data, pro­cesses and exchanges, so as to focus secu­ri­ty efforts on the most cri­ti­cal points. To do this, the deli­ve­rables of the EBIOS (Expres­sion des Besoins et Iden­ti­fi­ca­tion des Objec­tifs de Sécu­ri­té – Expres­sion of Needs and Iden­ti­fi­ca­tion of Secu­ri­ty Objec­tives) method should be used for the entire infor­ma­tion sys­tem under consi­de­ra­tion⁹.

Post-quan­tum risk assess­ment requires a glo­bal and coherent approach. It is vital to adapt exis­ting risk ana­lyses, which are based on the EBIOS method, to incor­po­rate the new threats posed by the emer­gence of quan­tum com­pu­ting. It is also essen­tial to coor­di­nate the ana­lyses car­ried out in iso­la­tion for each sys­tem or appli­ca­tion. An aggre­ga­tion phase is the­re­fore neces­sa­ry to obtain a sys­te­mic view of the risks and define a glo­bal secu­ri­ty stra­te­gy. This approach makes it pos­sible to iden­ti­fy the inter­de­pen­den­cies bet­ween the various ele­ments of the infor­ma­tion sys­tem and to put in place appro­priate pro­tec­tion measures.

The EBIOS method makes it pos­sible to iden­ti­fy not only the risks but also the sen­si­ti­vi­ty of the data held, which is cru­cial if the infor­ma­tion sys­tem is to be pro­tec­ted. It is essen­tial to dis­tin­guish bet­ween cri­ti­cal data, the loss of which could have a vital impact, and data that can be com­pro­mi­sed with fewer conse­quences. This assess­ment will make it pos­sible to prio­ri­tise pro­tec­tion efforts and ensure that secu­ri­ty mea­sures are adap­ted to the level of sen­si­ti­vi­ty of each type of data.

Consi­de­ra­tion should also be given to the pos­si­bi­li­ty of adap­ting secu­ri­ty pro­ce­dures by rever­ting to phy­si­cal means, such as paper, if neces­sa­ry. This approach can enhance the pro­tec­tion of sen­si­tive infor­ma­tion, par­ti­cu­lar­ly in high-risk situa­tions. The use of single-use codes is also a pro­mi­sing stra­te­gy. These codes, which have an expi­ry date of one to two months, can enhance secu­ri­ty by limi­ting the pos­si­bi­li­ties of unau­tho­ri­sed access.

In conclu­sion, the quan­tum threat repre­sents a major chal­lenge for the digi­tal secu­ri­ty of govern­ments and ope­ra­tors of vital impor­tance (ener­gy, com­mu­ni­ca­tions, trans­port, power, etc.). To meet these threats, it is essen­tial to anti­ci­pate them, invest in inno­va­tive solu­tions, mobi­lise human and finan­cial resources and pro­mote inter­na­tio­nal cooperation.

Quan­tum com­pu­ting will soon be capable of neu­tra­li­sing cur­rent encryp­tion sys­tems in record time, which means that we need to start adap­ting qui­ck­ly and tho­rough­ly right now. To achieve this, the tran­si­tion to post-quan­tum cryp­to­gra­phy will need to be cou­pled with robust secu­ri­ty mea­sures and increa­sed awa­re­ness, which are neces­sa­ry to gua­ran­tee our digi­tal sovereignty.

Final­ly, it is vital to take a glo­bal approach to the secu­ri­ty of digi­tal ser­vices, inte­gra­ting the tech­no­lo­gi­cal, sys­te­mic and human dimen­sions. The emer­gence of quan­tum tech­no­lo­gies is a major chal­lenge that needs to be addres­sed along­side cur­rent threats.