Mündges, S ephan; Pa k, Ki s y
A icle
Bu did hey eally? Pla o ms' compliance wi h he code o
p ac ice on disin o ma ion in e iew
In e ne Policy Re iew
P o ided in Coope a ion wi h:
Alexande on Humbold Ins i u e o In e ne and Socie y (HIIG), Be lin
Sugges ed Ci a ion: Mündges, S ephan; Pa k, Ki s y (2024) : Bu did hey eally? Pla o ms'
compliance wi h he code o p ac ice on disin o ma ion in e iew, In e ne Policy Re iew, ISSN
2197-6775, Alexande on Humbold Ins i u e o In e ne and Socie y, Be lin, Vol. 13, Iss. 3, pp.
1-21,
h ps://doi.o g/10.14763/2024.3.1786
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/300750
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/3.0/de/legalcode
Volume 13 |
Bu did hey eally? Pla o ms’ compliance
wi h he Code o P ac ice on
Disin o ma ion in e iew
S ephan Mündges TU Do mund Uni e si y
Ki s y Pa k Eu opean Digi al Media Obse a o y
DOI: h ps://doi.o g/10.14763/2024.3.1786
Published: 25 July 2024
Recei ed: 28 Decembe 2023 Accep ed: 27 Ma ch 2024
Funding: Ki s y Pa k ecei ed unding o his esea ch om he Eu opean Union
unde ac ion numbe 2020-EU-IA-0282 and ag eemen numbe INEA/CEF/ICT/A2020/
2381686, S ephan Mündges ecei ed unding o his esea ch om he Eu opean
Union, p ojec numbe 101083573.
Compe ing In e es s: The au ho has decla ed ha no compe ing in e es s exis ha
ha e in luenced he ex .
Licence: This is an open-access a icle dis ibu ed unde he e ms o he C ea i e
Commons A ibu ion 3.0 License (Ge many) which pe mi s un es ic ed use,
dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly
ci ed. h ps://c ea i ecommons.o g/licenses/by/3.0/de/deed.en
Copy igh emains wi h he au ho (s).
Ci a ion: Mündges, S. & Pa k, K. (2024). Bu did hey eally? Pla o ms’ compliance
wi h he Code o P ac ice on Disin o ma ion in e iew. In e ne Policy Re iew, 13(3).
h ps://doi.o g/10.14763/2024.3.1786
Keywo ds: Disin o ma ion, Pla o m egula ion, Code o P ac ice on Disin o ma ion, EU
in e ne policy, VLOP
Abs ac : A key pilla in he EU’s app oach o egula ing disin o ma ion is he S eng hened Code o
P ac ice on Disin o ma ion. This sel - egula o y Code p oposes a b oad ange o measu es o
di e en s akeholde s. I has been signed by pla o m companies ha he eby ag eed o epo on
compliance wi h he Code. This s udy in es iga es Google’s, Me a’s, Mic oso ’s, TikTok’s and
Twi e ’s (now X) compliance wi h hei epo ing obliga ions o he i s ime. Analysing he
pla o m baseline epo s published in ea ly 2023, we ind ha , o e all, pla o ms a e only pa ly
complian wi h he Code. Quali a i e in o ma ion p o ided by pla o ms o en lack de ail and/o
ele ance. Repo ed quan i a i e da a is, in se e al cases, missing, incomple e, o no obus . We
poin ou claims by pla o ms ha a e doub ul o ha e been p o en w ong in he pas , and
highligh a enues o u u e esea ch and in es iga ions. Addi ionally, we e lec on he amewo k
in place o moni o ing he Code o P ac ice and ways o imp o e i . This s udy is pa icula ly
ele an as he EU is ansi ioning om a sel - egula o y o a co- egula o y model when egula ing
disin o ma ion. The Code o P ac ice may soon become a code o conduc unde he Digi al Se ices
Ac (DSA) making non-compliance wi h i sanc ionable and inc easing he need o sys ema ic
moni o ing.
Issue 3
In oduc ion
Dis- and misin o ma ion is conside ed by many schola s, jou nalis s, and policy-
make s o be a subs an ial h ea o democ a ic sys ems and public discou se
(Lewandowsky e al., 2017; an de Linden, 2023). The e is deba e wi hin acade-
mia ega ding he scope o his challenge, wi h some schola s no ing ha empi i-
cal e idence o he p e alence o mis- and disin o ma ion and how i in luences
public a i udes, beha iou s, and pa icipa ion in poli ical p ocesses is con en ious
(Al ay e al., 2023; Al ay e al., 2022; Junghe & Sch oede , 2021). Howe e , o he
indings indica e ha alse claims and misin o ma ion a e indeed qui e p e alen ,
pa icula ly, on digi al pla o ms (Yang e al., 2023; Ecke e al., 2024). Indepen-
den o his academic deba e, se e al poli ical ac o s ha e aken ac ion o coun e
mis- and disin o ma ion, hinde i s sp ead in digi al in o ma ion spaces and – in
some ins ances – penalise ac o s publishing disin o ma ion (Cipe s e al., 2023).
Among democ a ic poli ical sys ems, he Eu opean Union is a guably he mos
p ominen and mos ac i e in his ega d. Apa om non-legisla i e ac ions (e.g.
mo e s a egic communica ion, mo e unding o media li e acy p og ammes), i
has ini ia ed he S eng hened Code o P ac ice on Disin o ma ion (he eina e e-
e ed o as ‘ he Code’ o ‘ he CoP’), a sel - egula o y amewo k signed by he ma-
jo echnology companies (Google, Me a, Mic oso , TikTok), among o he s (CoP,
2022). By becoming signa o ies, pla o ms commi o ake nume ous e y speci ic
ac ions o coun e mis- and disin o ma ion on hei se ices.
The Code is de ailed and wide- anging, co e ing ields such as poli ical ad e is-
ing, ecommenda ion sys ems, and suppo o esea che s and ac -checke s. Thus,
he epo ing ha he signa o ies o he Code publish biannually is ex ensi e. E en
hough only a olun a y commi men by he pla o ms, he Code is a egula o y in-
s umen which may help o p o ide anspa ency abou disin o ma ion on la ge
online pla o ms, bu he ex en o his anspa ency equi es sys ema ic moni o -
ing. This s udy summa ises da a and indings o such a moni o ing, ha was ca -
ied ou as a pan-Eu opean collabo a ion by independen academic esea che s
and has been epo ed o s akeholde s (Pa k & Mündges, 2023). Assessmen s o
signa o ies’ compliance wi h ce ain pa s o he Code ha e been conduc ed by
o he s akeholde s (EFCSN, 2023; Kiely e al., 2023), bu , o ou knowledge, his is
he i s comp ehensi e moni o ing o he Code’s pla o m signa o ies. Fo his, we
de eloped a moni o ing amewo k which we applied o he i s se o signa o y
epo s which we e published in ea ly 2023. Ou analysis p o ides insigh s in o
pla o ms’ compliance wi h he Code bu also o e s aluable lessons o u u e
moni o ing o simila policy ins umen s and may be ins uc i e o ongoing o mal
2 In e ne Policy Re iew 13(3) | 2024
p oceedings agains VLOPs. This is pa icula ly ele an as he Code o P ac ice on
Disin o ma ion migh become a Code o Conduc , and hence u n in o a legal im-
pe a i e, unde he Digi al Se ices Ac (DSA).
The Code o P ac ice on Disin o ma ion as a
co ne s one o he EU’s esponse o disin o ma ion
Disin o ma ion is a complex issue wi h a ying de ini ions and he need o bal-
ancing di e en undamen al igh s, making i challenging o egula e (Peuke ,
2023). S asi and Pa cu (2021) o e h ee concep ual models ha can be applied
when egula ing disin o ma ion:
TABLE 1: Regula o y models, adap ed om S asi & Pa cu, 2021
MODEL DESCRIPTION
STATE
INFLUENCE
LEVEL
WEAKNESSES
S a u o y
model
Rules passed by lawmake s o egula o y bodies,
en o ced by he go e nmen , go e nmen agencies,
o egula o y au ho i ies
s ong
Unlikely o keep up wi h as
changing echnology landscape;
could be used o jus i y censo ship
Co-
egula o y
model
Rules which a e nego ia ed by hose subjec o hem
and he egula o medium Risk o egula o y cap u e
Sel -
egula o y
model
Rules elying en i ely on olun a y compliance,
whe e legisla o s and egula o s mainly obse e
p i a e e o s and essen ially play no ac i e ole
weak
Shi o powe om public o p i a e
sec o ; lack o accoun abili y; lack o
p o ec ion o undamen al igh s
These models a e, o cou se, no ully dis inc om one ano he bu should be
hough o as a “spec um o possibili y” (S asi & Pa cu, 2021, p. 422). In he case
o egula ing disin o ma ion, he EU s a ed ou wi h he sel - egula o y model,
adop ing a mul i-s akeholde app oach, bu is shi ing i s app oach o e ecen
yea s o he co- egula o y and in pa s a u o y model.
The Code was ini ia ed by he Eu opean Commission and published in 2018, ma k-
ing he i s ime ha pla o ms ag eed on a se o sel - egula o y s anda ds (Eu o-
pean Commission, 2018). The ex o he Code was in o med by he epo o he
High-Le el G oup on ake news and online disin o ma ion and signed by se e al
s akeholde s including ad e ising associa ions and pla o m companies. A e i s
elease, he Code unde wen moni o ing by he Eu opean Commission i sel , as
well as by a ious s akeholde s, all o which concluded simila poin s o c i ique:
he Code was ague conce ning ce ain commi men s and insu icien ly comp e-
hensi e in o he s. Mo eo e , i was no ed ha he Code lacked measu able objec-
i es, key pe o mance indica o s, and a lack o independen e i ica ion o pla -
3 Mündges, Pa k
o m claims (Cullo y e al., 2021; Eu opean Regula o s G oup o Audio isual Media
Se ices, 2020; S asi & Pa cu, 2021). As a esul , i was e ised and e-published as
“S eng hened Code o P ac ice on Disin o ma ion” in 2022 (CoP, 2022). This new
i e a ion inco po a es nume ous sugges ions p e iously pu o h by s akeholde s
and moni o ing bodies, is be e s uc u ed, mo e de ailed, and clea e in wha in-
o ma ion mus be p o ided. Addi ionally, signa o ies a e obliged o epo mo e
g anula and coun y-speci ic da a.
The S eng hened CoP de ines disin o ma ion b oadly o encompass "misin o ma-
ion, disin o ma ion, in o ma ion in luence ope a ions, and o eign in e e ence in
he in o ma ion space" (CoP, 2022, p. 1). I consis s o nine sec ions, which can be
di ided in o h ee ca ego ies: eac ion, empowe men , and moni o ing.
• Reac ion: add esses mone isa ion o disin o ma ion and manipula i e
echniques.
◦ Sec ions: sc u iny o ad placemen s, poli ical ad e ising, in eg i y
o Se ices.
• Empowe men : manda es coope a ion wi h s akeholde s.
◦ Sec ions: empowe ing use s, esea ch communi y, ac -checking
communi y.
• Moni o ing: de ails how he Code is o e seen and i s ou comes
dissemina ed.
◦ Sec ions: anspa ency cen e, pe manen ask- o ce, moni o ing o
he Code.
Each sec ion has h ee laye s:
1. Commi men s: o e all objec i es and ac ions
2. Measu es: speci ic s eps o Signa o ies
3. Repo ing elemen s: ei he "Quali a i e Repo ing Elemen s" (QRE) o
quan i iable "Se ice Le el Indica o s" (SLI); hey de ail wha signa o ies
ha e o epo in o de o p o e compliance wi h measu es
Fo example, commi men 27 obliges pla o ms “ o p o ide e ed esea che s
wi h access o da a necessa y o unde ake esea ch on Disin o ma ion by de elop-
ing, unding, and coope a ing wi h an independen , hi d-pa y body ha can e
esea che s and esea ch p oposals” which is ope a ionalized in ou measu es: o
con ibu e o he de elopmen o he a o emen ioned independen hi d-pa y
body (measu e 27.1); commi o co- und his body (27.2): wo k wi h his hi d-pa y
body “ o enable sha ing o pe sonal da a necessa y o unde ake esea ch on Dis-
in o ma ion wi h e ed esea che s” (27.3); engage in pilo p og ams o g an ing
da a access o esea che s wi hou he hi d-pa y body being se up (27.4). Fo
4 In e ne Policy Re iew 13(3) | 2024
each measu e signa o ies ha e o epo quali a i e in o ma ion, speci ied in he
co esponding QREs (e.g. QRE 27.2.1: disclose unding o he de elopmen o he
hi d-pa y body), and o measu e 27.3 hey also ha e o epo quan i a i e da a
abou he numbe o esea ch p ojec s which ha e been g an ed da a access.
Cu en ly, he Code includes 44 commi men s which a e ope a ionalized in 128
measu es. I has been signed by 44 s akeholde s, including big ech companies
Google, Me a, Mic oso , and TikTok. Twi e (now called X; as he pla o m com-
mi ed o he Code as Twi e his name is used h oughou his pape ) had been a
signa o y un il May 2023 bu he company decided o lea e he Code unde i s
new owne Elon Musk. Gi en i s design, i is e iden he CoP p ima ily aims o eg-
ula e Big Tech's ole in disin o ma ion dissemina ion. By signing he Code, signa o-
ies commi o epo egula ly, e e y six mon hs in he case o signa o ies ha
quali y as Ve y La ge Online Pla o ms (VLOPs) o Sea ch Engines (VLOSEs) unde
he DSA, on hei compliance wi h he Code. They do so using a p ede ined epo -
ing empla e which enables bo h quan i a i e as well as quali a i e compa isons
be ween signa o ies. The empla e was de eloped by ep esen a i es o he Eu o-
pean Regula o s G oup o Audio isual Media Se ices (ERGA) ha a e pa o he
CoP Pe manen Task- o ce in consul a ion wi h he signa o ies.
Pa icipa ion in he Code is olun a y. Howe e , he DSA will in oduce obliga ions
o VLOPs and VLOSEs de ined as pla o ms wi h mo e han 45 million use s in he
EU. The DSA's A icle 45 commends he o mula ion o olun a y codes o conduc ,
e e encing he CoP as a po en ial example (Regula ion 2022/2065, 106). Con-
c e ely, a icle 45(1) asks he Commission and he Eu opean Boa d o Digi al Se -
ices wi h p omo ing and acili a ing he c ea ion o codes o conduc o ensu e
p ope applica ion o he DSA, pa icula ly add essing he challenges o sys emic
isks. A icle 45(2) s a es ha i signi ican sys emic isks in ol ing se e al VLOPs
and/o VLOSEs a ise, he Commission may in i e s akeholde s o de elop codes
wi h speci ic isk mi iga ion measu es and a epo ing amewo k.
Hence, codes o conduc a e pa icula ly ele an o VLOPs and VLOSEs in ul ill-
ing wo co ne s one a icles o he DSA, namely a icles 34 “Risk assessmen ” and
35 “Mi iga ion o isks”. Risk assessmen s mus encompass a ious issues. Al hough
he a icle does no explici ly men ion disin o ma ion, i implies comp ehensi e
co e age o he phenomenon by including “any ac ual o o eseeable nega i e e -
ec s on ci ic discou se and elec o al p ocesses, and public secu i y” (A . 34(1)(c)).
Mi iga ion measu es mus be aken by VLOPs and VLOSEs i sys emic isks a e
iden i ied in he assessmen s conduc ed as pa o a icle 34. These measu es may
include modi ying se ice designs and ea u es, upda ing e ms and condi ions, im-
5 Mündges, Pa k
p o ing con en mode a ion p ocesses o quickly add ess illegal con en , o adjus -
ing “in e nal p ocesses, esou ces, es ing, documen a ion, o supe ision o any o
hei ac i i ies in pa icula as ega ds de ec ion o sys emic isk” (A . 35(1)( )).
Codes o conduc o e he oppo uni y o speci y hese b oad obliga ions and op-
e a ionalise hem in o speci ic commi men s. Addi ionally, hey may p o ide de-
ailed s anda ds o acili a e in e nal compliance and ex e nal assessmen s, gi e a
di e se se o s akeholde s he chance o shape DSA implemen a ion and allow o
u u e ac ion and adap a ion o echnological p og ess by add essing a eas o gaps
no co e ed in he DSA.
While hese codes a e so law ins umen s and will con inue o be olun a y, he
DSA does speci y ha non-compliance wi h codes may a ac penal ies, which is a
much bigge incen i e o majo pla o ms o coope a e (G i in & Vande Maelen,
2023) and signals he EU’s shi om a sel - o a co- egula o y model when egu-
la ing disin o ma ion. As a p obable inaugu al code unde he DSA, i is c ucial o
he CoP o be unambiguous, s ic ly adhe ed o, and consis en ly moni o ed.
Conside ing he b oad scope o he CoP, i s ocus on e y la ge digi al pla o ms
and he ele ance o moni o ing compliance sys ema ically, we ask wo esea ch
ques ions:
RQ1: To wha ex en do pla o m signa o ies o he Code comply wi h hei epo -
ing obliga ions?
RQ2: How can compliance wi h he CoP be moni o ed sys ema ically, egula ly and
comp ehensi ely?
Me hod
This analysis cen es on pla o m signa o ies classi ied as VLOPs/VLOSEs unde he
DSA, which a e Google, Me a, Mic oso , TikTok, and Twi e . These pla o ms a e
o en exploi ed by malicious en i ies and unsuspec ing use s o dissemina e disin-
o ma ion and misin o ma ion. Gi en hei as use bases, hey wa an signi ican
sc u iny. To answe RQ1 we analysed he i s se o CoP epo s published by he
pla o ms in ea ly 2023. The epo s a e ex ensi e (859 pages in o al) bu he
s uc u e o he p ede e mined epo ing empla e enabled obus compa isons o
signa o y epo s. We op ed o analyse a he measu e le el, iewing commi men s
as oo o e a ching and epo ing elemen s as o e ly g anula .
In ou analysis we d ew om con en analysis me hodology (Puppis, 2019). We o -
6 In e ne Policy Re iew 13(3) | 2024
mula ed a cus omised coding scheme, inco po a ing h ee quan i a i e and wo
quali a i e me ics, de ailed in he subsequen able.
TABLE 2: Coding scheme
VARIABLE EXPRESSION EXPLANATION
MISSING
QUALITATIVE
RESPONSES
yes
In o ma ion ha is eques ed in a Measu e o QRE is no p o ided comp ehensi ely,
and/o he in o ma ion p o ided by he Signa o y does no mee he obliga ions se ou
in he Measu e o QRE. As some QREs co e se e al aspec s o an issue, his a iable
should also be coded as “yes” when he Signa o y ails o p o ide in o ma ion o one o
mo e ele an aspec s.
no The quali a i e esponses p o ided by he Signa o y a e comp ehensi e, concise and o
he poin .
i ele an I a Measu e does no con ain QREs, his a iable migh be coded i ele an .
MISSING
QUANTITATIVE
DATA
yes
SLIs a e igno ed o ele an da a ha is eques ed as pa o a Measu e o SLI is no
p o ided by he Signa o y. This migh also be he case i Membe S a e speci ic da a is
eques ed bu no p o ided.
The a iable is also o be coded “yes” i da a p o ided is unsui able o he me hodology
is ill-sui ed o p o ide eliable da a.
no Quan i a i e da a p o ided by he Signa o y is comp ehensi e and he me hodology o
collec o calcula e he da a is app op ia e.
i ele an Measu e does no con ain SLI o he e is easonable g ound why an SLI canno be
p o ided a his poin in ime.
OVERALL
SCORE
1
Poo : he esponse signi ican ly alls sho o mee ing he equi emen s o he Measu e.
This is he case o esponses ha lack majo de ails, a e incomple e o i ele an , o
ail o add ess he speci ic in o ma ion eques s ou lined in he measu e.
2
Adequa e: he esponse shows e o owa ds mee ing he equi emen s o he Measu e
bu he e a e no able issues o a eas ha equi e imp o emen . This is he case o
esponses ha pa ially add ess he ques ion, bu may lack impo an de ails, e idence,
o con ex .
3
Good: he esponse ully mee s he equi emen s o he Measu e. This is he case o
esponses ha a e comple e, ele an , and p o ide clea and comp ehensi e
in o ma ion ha di ec ly add esses he speci ic in o ma ion eques s ou lined in he
measu e.
n/a
No Applicable: i a Signa o y claims a Measu e hey subsc ibed o is no ele an o
hei se ices and he Assesso belie es his claim o be co ec e.g. he Measu e ela es
o displaying in o ma ion alongside poli ical ad e ising and he Signa o y's p oduc
does no allow poli ical ad e ising.
SIGNATORY
LEVEL
COMMENTS
ee ex
Quali a i e commen s explaining he Code ’s esul s o he p e ious h ee a iables,
e.g. “lacked de ail on x”. Code s should include any obse a ions o eedback on he
esponse.
CODE LEVEL
COMMENTS ee ex Quali a i e commen s ega ding he CoP; his migh include commen s on wo ding,
s uc u e, needed cla i ica ion, e c.
Fo coding, we adop ed a h ee-s ep p ocess. Ini ially, wo code s independen ly
analysed speci ic sec ions, wi h he op ion o lag unce ain codings as "needs a
second opinion". A e hese ini ial ounds, a hi d ound sough o inalise he e-
sul s. The lead au ho s e iewed all disc epancies be ween he ini ial code s, e e -
7 Mündges, Pa k
encing commen s and epo s con en o de e mine he inal ou come. Addi ional-
ly, hey e iewed all codings o inconsis encies.
We p esen nume ical ou comes o he i s six sec ions. Fo he inal h ee, o-
cused on he Code's moni o ing me hods, we o e gene al ema ks wi hou nume -
ical da a, as many measu es lacked dis inc epo ing elemen s o we e absen en-
i ely. F om ou quali a i e analysis o he las h ee sec ions as well as om in-
sigh s gained h ough he coding p ocess o he i s six sec ions we d aw conclu-
sions o o e po en ial answe s o RQ2.
Resul s o RQ1: Assessmen o pla o ms’ compliance
wi h he Code o P ac ice on Disin o ma ion
TABLE 3: O e all sco es pe signa o y (sco es co espond o g ades o 1 = Poo , 2 = Adequa e and 3
= Good)
SIGNATORIES % OF MEASURES MISSING QUALITATIVE
INFORMATION
% OF MEASURES MISSING
QUANTITATIVE DATA
AVERAGE OVERALL
SCORE
Google 44% 53% 2.1
Me a 52% 65% 2.0
Mic oso 54% 59% 1.9
TikTok 43% 50% 2.0
TOTAL w/o
Twi e 49% 58% 2.0
Twi e 100% 100% 1.0
TOTAL 55% 64% 1.9
The a e age sco e o signa o ies is a subpa 1.9, wi h only Google exceeding an
o e all g ade o 2 (Adequa e) a 2.1. Measu es con aining QREs e ealed ha 55%
(wi h Twi e excluded 49%) we e ei he incomple e o lacked quali a i e in o ma-
ion, while 64% (wi h Twi e excluded 58%) o Measu es wi h SLIs we e lacking a
leas some da a.
Twi e le he Code in May 2023, so se e al mon hs a e he baseline epo s had
been published (Hend ix, 2023). E idence sugges s i s commi men was waning
well be o e i s exi , a e aging he lowes possible sco e 1 (Poo ) and exhibi ing a
100% de iciency in QREs/SLIs ac oss all measu es. The epo con ained nume ous
emp y sec ions, limi ed gene al commen s on some commi men s, and in o ma ion
ha o en seemed di ec ly copied om hei websi e. This sugges s Twi e had li -
le in en ion o adhe ing o he Code when epo ing.
8 In e ne Policy Re iew 13(3) | 2024
Twi e unde he DSA ci ing sho comings o “measu es aken o comba in o ma-
ion manipula ion on he pla o m” (Eu opean Commission, 2023).
Rega ding he emaining pla o ms, he in o ma ion and esponses p o ided o en
ail o p ecisely add ess he Code's speci ic equi emen s, wi h many ins ances o
incomple e in o ma ion. The epo ing o quan i a i e da a equi es signi ican im-
p o emen in bo h scope and de ail. The me hodologies o calcula ing quan i a-
i e da a need o be mo e accu a e, wi h an emphasis on p o iding membe s a e-
le el da a.
I is c ucial o emphasise ha ou analysis mus be egula ly eplica ed and e ined
o maximise i s u ili y o policy e alua ion and en o cemen . We encou age e-
sea che s o build upon and apply ou me hodological amewo k o o he pla -
o m epo s on compliance wi h he CoP. We belie e i is easily eplicable and
pla o m-agnos ic, i.e. i mo e VLOPs o VLOSEs join he Code hei epo s can be
analysed applying he same me hodological amewo k.
Signa o ies and egula o s should pay special a en ion o ensu ing compa able
me hodologies when epo ing SLIs. Al hough challenging, due o he pla o ms'
di e ing ocuses, me ics, and speci ica ions ha complica e compa isons, he e
a e s ill da a poin s ha allow o compa isons (e.g. ela ed o TTPs). Addi ionally,
he da a cu en ly hidden in pla o m epo s could be in eg a ed in o he DSA
T anspa ency Da abase2 and made a ailable in a machine- eadable o ma .
Repea ed esea ch o e ime will yield in aluable insigh s in o how pla o ms
e ol e in coun e ing disin o ma ion, how he implemen a ion o he DSA in lu-
ences hei policies and en o cemen ac ions, and hei o e all beha iou . This i e -
a i e p ocess will e ine he me hodology, iden i y gaps in pla o m epo ing, and
highligh s a emen s ha wa an u he in es iga ion. This will be pa icula ly
signi ican when he Code o P ac ice ansi ions in o a code o conduc unde he
DSA. As i is likely o be he i s one, he way his ansi ion is managed, i s im-
pac , and i s implemen a ion by egula o s and policymake s will la gely de e -
mine he DSA's e ec i eness in egula ing ha m ul bu mos ly law ul speech.
Gi en ecen de elopmen s and in es iga ions ini ia ed by he Eu opean Commis-
sion in o he beha iou and policies o e y la ge online pla o ms, he impo ance
o his wo k canno be o e s a ed. In i s p ess elease announcing he opening o
o mal p oceedings agains Me a, he Eu opean Commission men ions speci ically
2. See Eu opean Commission (n.d.).
15 Mündges, Pa k
ha i “suspec s ha Me a does no comply wi h DSA obliga ions ela ed o ad-
d essing he dissemina ion o decep i e ad e isemen s, disin o ma ion campaigns
and coo dina ed inau hen ic beha iou in he EU” (Eu opean Commission, 2024).
Moni o ing compliance wi h he Code o P ac ice may o e aluable insigh s o
such in es iga ions and indica e a eas whe e pla o ms all sho in hei esponsi-
bili y o coun e disin o ma ion. E en hough ou analysis was ocussed on moni-
o ing compliance wi h epo ing obliga ions, we ound se e al ins ances in which
pla o ms’ claims a e a leas doub ul i no e en con adic ed by al eady pub-
lished empi ical e idence. A eas o conce n a e he en o cemen o bans on poli i-
cal ad e isemen s and ac ions agains ake engagemen and inau hen ic beha -
iou . The e o e, u u e moni o ing should be accompanied by in-dep h in es iga-
ions in o speci ic ocus a eas o u he enhance ou unde s anding and imp o e
egula o y p ac ices. The Eu opean Commission should ensu e his moni o ing is
adequa ely unded and esou ced long- e m, conduc ed by independen e-
sea che s wi h he necessa y backg ound knowledge. Only a e hese s eps o-
wa ds a sys ema ic moni o ing ha e been aken, will we be able o answe wi h
ce ain y he ques ion: bu did pla o ms eally do wha hey claim o ha e done?
Limi a ions
While his s udy p o ides aluable insigh s in o pla o ms’ compliance wi h he
CoP, i is essen ial o acknowledge ce ain limi a ions. Fi s ly, ou analysis is based
on he ini ial se o epo s published by pla o m signa o ies in ea ly 2023. Since
hen, hese signa o ies ha e published a second ound o epo s in Sep embe
2023. Ou esul s do no e lec any changes o epo ing compliance h ough
hese epo s. Secondly, he moni o ing amewo k de eloped o his analysis,
while comp ehensi e, may ha e inhe en subjec i i y in i s assessmen c i e ia.
The in e p e a ion o policy compliance and epo ing quali y can a y. Th ough
applica ion o a h ee-s ep coding p ocess we y o minimise his subjec i i y bu
may no elimina e i en i ely. Thi dly, he analysis is ocused on moni o ing compli-
ance wi h epo ing obliga ions. I does no comp ehensi ely assess he accu acy
o in o ma ion and da a epo ed by signa o ies. As no ed, we iden i ied se e al in-
s ances in which accu acy is a leas doub ul. This calls o mo e in es iga o y ap-
p oaches o moni o CoP compliance. Las ly, he analysis ocuses on pla o m sig-
na o ies, and he indings may no gene alise o pla o ms ha a e no (ye ) signa-
o ies o he Code.
16 In e ne Policy Re iew 13(3) | 2024
ACKNOWLEDGEMENTS
The au ho s hank all colleagues who con ibu ed o he analysis o pla o m
epo s and ga e eedback du ing he d a ing o he a icle:
Ma o B au o ić, Uni e si y o Dub o nik
Eileen Cullo y, Dublin Ci y Uni e si y
Dawn Hol o d, Uni e si y o B is ol
Anas asia Kozy e a, Max Planck Ins i u e o Human De elopmen
S ephan Lewandowsky, Uni e si y o B is ol, Uni e si y o Po sdam, and Uni e si y
o Wes e n Aus alia
T isha Meye , V ije Uni e si ei B ussel
Susanne Wegne , TU Do mund Uni e si y
Ki s y Pa k and Eillen Cullo y ecei ed co- unding o his esea ch om he
Eu opean Union unde ac ion numbe 2020-EU-IA-0282 and ag eemen numbe
INEA/CEF/ICT/A2020/2381686.
Susanne Wegne and S ephan Mündges ecei ed co- unding o his esea ch om
he Eu opean Union, p ojec numbe 101083573.
Ma o B au o ić ecei ed co- unding o his esea ch om he Eu opean Union,
p ojec numbe 101083909.
T isha Meye ecei ed co- unding o his esea ch om he Eu opean Union,
p ojec numbe INEA/CEF/ICT/A2020/2394296.
S ephan Lewandowsky has ecei ed unding om Jigsaw (a echnology incuba o
c ea ed by Google) o se e al empi ical p ojec s un ela ed o he p esen pape .
He also acknowledges inancial suppo om he Eu opean Resea ch Council (ERC
Ad anced G an 101020961 PRODEMINFO) and he Eu opean Commission
(Ho izon 2020 g an s 964728 JITSUVAX and 101094752 SoMe4Dem). He also
ecei es suppo om UK Resea ch and Inno a ion ( h ough EU Ho izon
eplacemen unding g an numbe 10049415).
Anas asia Kozy e a and S ephan Lewandowsky acknowledge unding om he
Volkswagen Founda ion (g an “Reclaiming indi idual au onomy and democ a ic
discou se online: How o ebalance human and algo i hmic decision making”).
17 Mündges, Pa k
Re e ences
Al ay, S., Be iche, M., & Ace bi, A. (2023). Misin o ma ion on misin o ma ion: Con-
cep ual and me hodological challenges. Social Media + Socie y, 9(1). h ps://doi.o g/
10.1177/20563051221150412
Al ay, S., Kleis Nielsen, R., & Fle che , R. (2022). Quan i ying he “in odemic”: Peo-
ple u ned o us wo hy news ou le s du ing he 2020 co ona i us pandemic.
Jou nal o Quan i a i e Desc ip ion: Digi al Media, 2. h ps://doi.o g/10.51685/
jqd.2022.020
Cipe s, S., Meye , T., & Le e e e, J. (2023). Go e nmen esponses o online disin o -
ma ion unpacked. In e ne Policy Re iew, 12(4). h ps://doi.o g/10.14763/
2023.4.1736
Cullo y, E., Pa k, K., Feenane, T., Papae angelou, C., Con oy, A., & Sui e , J. (2021).
Co idcheck: Assessing he implemen a ion o EU code o p ac ice on disin o ma ion in
ela ion o Co id-19 [P ojec epo ]. DCU Ins i u e o Fu u e Media, Democ acy and
Socie y. h ps://do as.dcu.ie/26472/
Ecke , U., Roozenbeek, J., an de Linden, S., Tay, L. Q., Cook, J., O eskes, N., &
Lewandowsky, S. (2024). Misin o ma ion poses a bigge h ea o democ acy han
you migh hink. Na u e, 630(8015), 29–32. h ps://doi.o g/10.1038/
d41586-024-01587-3
Eu opean Commission. (2018). Tackling online disin o ma ion: A Eu opean app oach
(Communica ion COM(2018) 236 inal). h ps://eu -lex.eu opa.eu/legal-con en /EN/
TXT/HTML/?u i=CELEX:52018DC0236
Eu opean Commission. (2023). Commission opens o mal p oceedings agains X unde
he Digi al Se ices Ac [P ess elease]. h ps://ec.eu opa.eu/commission/p ess-
co ne /de ail/en/IP_23_6709
Eu opean Commission. (2024). Commission opens o mal p oceedings unde DSA
[P ess elease]. h ps://ec.eu opa.eu/commission/p essco ne /de ail/en/ip_24_2373
Eu opean Commission. (n.d.). DSA anspa ency da abase. h ps:// anspa en-
cy.dsa.ec.eu opa.eu/
Eu opean Fac -Checking S anda ds Ne wo k. (2023). Fac -checke s’ eedback on he
18 In e ne Policy Re iew 13(3) | 2024
baseline epo s o he VLOP signa o ies o he Code o P ac ice on Disin o ma ion [Re-
po ].
Eu opean Regula o s G oup o Audio isual Media Se ices. (2020). ERGA Repo on
disin o ma ion: Assessmen o he implemen a ion o he Code o P ac ice [Repo ].
h ps://e ga-online.eu/wp-con en /uploads/2020/05/ERGA-2019- epo -pub-
lished-2020-LQ.pd
F edheim, R., Bay, S., Dek, A., S olze, M., & Haiduchyk, T. (2023). Social media ma-
nipula ion 2022/2023: Assessing he abili y o social media companies o comba pla -
o m manipula ion [Repo ]. NATO S a egic Communica ions Cen e o Excellence.
h ps://s a comcoe.o g/publica ions/social-media-manipula ion-20222023-as-
sessing- he-abili y-o -social-media-companies- o-comba -pla o m-manipula ion/
272
Google. (2023). Code o P ac ice on Disin o ma ion – Repo o Google o he pe iod 1
July 2022—30 Sep embe 2022 [Da ase ]. T anspa ency Cen e. h ps://disin-
ocode.eu/ epo s-a chi e/?yea s=2023
G i in, R., & Vande Maelen, C. (2023). Codes o conduc in he Digi al Se ices Ac :
Explo ing he oppo uni ies and challenges. SSRN. h ps://doi.o g/10.2139/ss-
n.4463874
Hend ix, J. (2023, May 26). Musk’s Twi e di ches EU Code o P ac ice on Disin o -
ma ion. Tech Policy P ess. h ps:// echpolicy.p ess/musks- wi e -di ches-eu-code-
o -p ac ice-on-disin o ma ion/
High Le el G oup on Fake News and Online Disin o ma ion. (2018). A mul i-dimen-
sional app oach o disin o ma ion [Repo ]. Eu opean Commision. h ps://ec.eu-
opa.eu/news oom/dae/documen .c m?doc_id=50271
Junghe , A., & Sch oede , R. (2021). Disin o ma ion and he s uc u al ans o ma-
ions o he public a ena: Add essing he ac ual challenges o democ acy. Social
Media + Socie y, 7(1). h ps://doi.o g/10.1177/2056305121988928
Kiely, K. P., Ma go a, R., Ga go a, S., Dob e a, M., Gando a, T., & S e ano a, T.
(2023). E alua ing VLOP and VLOSE implemen a ion o he S eng hened EU Code o
P ac ice on Disin o ma ion in Bulga ia [Whi e pape ]. GATE Ins i u e Big Da a o
Sma Socie y. h ps://b odhub.eu/en/ esea ch/e alua ing- lop-and- lose-imple-
men a ion-o - he-s eng hened-eu-code-o -p ac ice-on-disin o ma ion-in-bulga -
ia/
19 Mündges, Pa k
Lewandowsky, S., Ecke , U. K. H., & Cook, J. (2017). Beyond misin o ma ion: Unde -
s anding and coping wi h he “pos - u h” e a. Jou nal o Applied Resea ch in Memo-
y and Cogni ion, 6(4), 353–369. h ps://doi.o g/10.1016/j.ja mac.2017.07.008
Me a. (2023). Code o P ac ice on Disin o ma ion – Me a baseline epo [Da ase ].
T anspa ency Cen e. h ps://disin ocode.eu/ epo s-a chi e/?yea s=2023
Mic oso . (2023). Code o P ac ice on Disin o ma ion Baseline Repo – Janua y 2023
Mic oso [Da ase ]. T anspa ency Cen e. h ps://disin ocode.eu/ epo s-
a chi e/?yea s=2023
Mozilla Founda ion. (2021). Th€se a e no po£i ical ad$: How pa isan in luence s a e
e ading TikTok’s weak poli ical ad policies [Repo ]. h ps:// ounda ion.mozilla.o g/
documen s/178/TikTok-Ad e ising-Repo _e5G Wx5.pd
Nenadic, I., B ogi, E., & Bleye -Simon, K. (2023). S uc u al indica o s o assess e ec-
i eness o he EU’s Code o P ac ice on Disin o ma ion (Wo king Pape 2023/34). Eu-
opean Uni e si y Ins i u e. h ps://hdl.handle.ne /1814/75558
Pa k, K., & Mündges, S. (2023). CoP moni o : Baseline epo s. Assessmen o VLOP
and VLOSE Signa o y epo s o he S eng hened Code o P ac ice on Disin o ma ion
[Repo ]. EDMO I eland & Ge man-Aus ian Digi al Media Obse a o y. h ps://gad-
mo.eu/wp-con en /uploads/2023/09/CoP-Moni o -Repo .pd
Peuke , A. (2023). Desin o ma ions egulie ung in de EU – Übe blick und o ene
F agen [Disin o ma ion egula ion in he EU - o e iew and open ques ions]. Ju is-
enZei ung, 78(7), 278–296. h ps://doi.o g/10.1628/jz-2023-0095
Puppis, M. (2019). Analyzing alk and ex I: Quali a i e con en analysis. In H. Van
den Bulck, M. Puppis, K. Donde s, & L. Van Audenho e (Eds.), The Palg a e hand-
book o me hods o media policy esea ch (pp. 367–384). Palg a e Macmillan.
h ps://doi.o g/10.1007/978-3-030-16065-4_21
Regula ion 2022/2065. (2022). Regula ion (EU) 2022/2065 o he Eu opean Pa lia-
men and o he Council o 19 Oc obe 2022 on a Single Ma ke Fo Digi al Se ices
and amending Di ec i e 2000/31/EC (Digi al Se ices Ac ). Eu opean Pa liamen and
Council. h ps://eu -lex.eu opa.eu/eli/ eg/2022/2065
S asi, M. L., & Pa cu, P. L. (2021). Disin o ma ion and misin o ma ion: The EU e-
sponse. In P. L. Pa cu & E. B ogi (Eds.), Resea ch handbook on EU media law and poli-
cy. Edwa d Elga Publishing. h ps://doi.o g/10.4337/9781786439338.00030
20 In e ne Policy Re iew 13(3) | 2024
The S eng hened Code o P ac ice on Disin o ma ion. (2022). h ps://disin ocode.eu/
wp-con en /uploads/2023/01/The-S eng hened-Code-o -P ac ice-on-Disin o ma-
ion-2022.pd
TikTok. (2023). Code o P ac ice on Disin o ma ion – Repo o TikTok o he pe iod 16
June – 16 Decembe 2022 [Da ase ]. T anspa ency Cen e. h ps://disin ocode.eu/ e-
po s-a chi e/?yea s=2023
T us Lab. (2023). A compa a i e analysis o he p e alence and sou ces o disin o ma-
ion ac oss majo social media pla o ms in Poland, Slo akia, and Spain [Repo ].
h ps://my. isme.co/ iew/ dp xy4j-code-o -p ac ice-on-misin o ma ion-sep em-
be -2023#s1
an de Linden, S. (2023). Foolp oo : Why misin o ma ion in ec s ou minds and how
o build immuni y. W.W. No on & Company.
Visse , F., Smi no a, J., & Ma iny, C. (2023). Cashing in on con lic : TikTok p o i s
om p o-K emlin disin o ma ion ads (Digi al Dispa ches). Ins i u e o S a egic Dia-
logue. h ps://www.isdglobal.o g/digi al_dispa ches/cashing-in-on-con lic - ik ok-
p o i s- om-p o-k emlin-disin o ma ion-ads/
Yang, Y., Da is, T., & Hindman, M. (2023). Visual misin o ma ion on Facebook. Jou -
nal o Communica ion, 73(4), 316–328. h ps://doi.o g/10.1093/joc/jqac051
in coope a ion wi hPublished by
21 Mündges, Pa k
