FEBRUARY2025

InternationalCoordinationforAccountabilityinAIGovernance

KeyTakeawaysfromtheSixthEditionof theAthensRoundtableonArtificialIntelligenceandtheRuleofLawHeldDecember9,2024,inParis,France

GeorgeGor

,

FrankRyan

,

TerezaZoumpalova

,

NikiIliadis

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

TableofContents

AbouttheSixthEdition

oftheAthensRoundtableonArtificialIntelligenceandtheRuleofLaw

4

RecommendationsforKeyStakeholders

5

Governments

5

IntergovernmentalOrganizations

6

Industry

6

CivilSociety

6

Individuals

6

TheSixthEditionInNumbers

8

AgendaOverview

12

SessionTakeaways

13

OpeningRemarks

13

Session1:MappingtheLandscapeofAIGovernanceandAccountability

17

Session2:KeyChallengesforEstablishingAccountabilityInGovernance

21

Session3:BuildingStrongerAccountabilityAcrossActorsGlobally

25

Session4:KeyTakeawaysfromtheConsultationfortheAIActionSummit

29

ClosingRemarks

33

Acknowledgements

36

SummaryofKeyInsightsandConsiderations

37

2

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

TheSixthEditionoftheAthens

RoundtableonAIandtheRuleofLaw

OECDHeadquarters,Paris,FranceDecember9,2024

UnderthepatronageofH.E.thePresidentoftheHellenicRepublicMs.KaterinaSakellaropoulou

ORGANIZEDBY:

INPARTNERSHIPWITH:

WITHSUPPORTFROM:

THE

FUTURE

SOCIETY

Arnold&porter

3

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

AboutTheSixthEdition

TheSixthEditionoftheAthensRoundtableonAI

andtheRuleofLaw,hostedattheOECDinParisonDecember9,providedavaluableplatformforengagingwithoneofthedefiningchallengesofourtime:establishingrobustaccountability

mechanismstoensureAIisdevelopedandgovernedinalignmentwiththeruleoflaw.

Convening135in-personparticipants—includinggovernmentministers,policymakers,civilsocietyleaders,industryexecutives,andacademics—and830onlineattendeesfrom108countries,the

eventunderscoredtheurgencyofthistaskamidintensifyinggeopoliticalcompetitionandan

acceleratingracefortechnologicaldominance.

Wewerehonoredtohostthiseventunderthe

patronageofHerExcellencythePresidentoftheHellenicRepublic,Ms.KaterinaSakellaropoulou,andinpartnershipwiththeOECD,UNESCO,theAI&SocietyInstitute,CNNum,andM,withsupportfromArnold&PorterandFathom.

Significantly,thisyear’seditionofTheAthens

RoundtableservedasakeymilestoneforshapingtheagendaanddeliverablesoftheAIAction

SummitforthcominginFebruary2025.

Discussionsrevolvedaroundfourthemes:

assessingprogressinAIgovernance,identifyinggapsinexistinggovernanceregimes,crafting

solutions,anddefiningactionableprioritiesfortheAIActionSummit.Participantsstressedthe

importanceofinternationalcollaborationto

establishbindingglobalAIstandardsand

accountabilitymechanisms.Rigoroustesting,

independentevaluations,andthird-partyaudits,

supportedbyAISafetyInstitutes,wererecognizedasvitalforbuildingsafeandaccountableAI.WhileadvancingscientificconsensusonAIrisks

remainsessential,ourdiscussionshighlightedthatimmediategovernanceactionscannotwait.

Broadersocietalparticipation,especiallyfromcivilsociety,wasdeemedcrucialtoensuring

governanceframeworksreflectdiverseperspectivesandglobalneeds.

Aswereflectontheoutcomesofthisevent

throughoutthisreport,onethingisclear:buildingeffectivegovernanceforAIisnotanendpointbutanongoingendeavor.Thestakesarehigh:

withouteffectiveaccountabilitymechanisms,AIwillexacerbateinequality,erodehumanrights,anddestabilizeourdemocraticinstitutions.Ashighlightedinouropeningremarks:“ifAI

developersholdthemselvesaccountabletonobody,theywon’tbetrustedbyanybody.”

Urgently,weneedthecollectiveeffortof

thousandsofindividualsworldwideand

coordinationacrossgeographicregionsand

sectors.Thisishowwecanfacilitatethe

mitigationofsystemicrisksposedbyAIandtheresponsibleandequitablerealizationofits

promises—notonlyforourselves,butforthefutureofoursociety.

Withappreciationandhope,

NickMo?s

ExecutiveDirectorTheFutureSociety

NikiIliadis

Director,GlobalAIGovernanceTheFutureSociety

4

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

RecommendationsforKeyStakeholders

WhoholdstheresponsibilityforensuringthatthedevelopmentanddeploymentofAIserveto

benefithumanity?Ourviewisthatthismustbeasharedeffort,requiringproactiveengagement

fromgovernments,intergovernmental

organizations,industry,civilsociety,and

individuals.Eachactorbringsunique

perspectives,resources,andexpertisetothe

governanceprocess.Together,wecantacklethesignificantworkneededtoestablishrobust,

effective,andgloballyconsistentframeworks.

DrawingfromdiscussionsattheSixthEditionof

theAthensRoundtableonAIandtheRuleofLaw,

GOVERNMENTS

HarmonizeInternationalStandards

EngageinbilateralandmultilateraleffortstoalignAIgovernanceframeworksthroughestablished

internationalmechanismssuchastheOECDAI

Principles,ensuringinteroperabilityacross

jurisdictions.Securehigh-levelpolitical

commitmentfromheadsofstatetoestablishandmutuallyrecognizeinternationalstandards.ThisincludessupportingaParisDeclarationsignedattheAIActionSummitbyadiversecoalitionof

worldleaderstofoster“constructivevigilance”

andproactiveoversightinAIdevelopment.

Eachactorbringsuniqueperspectives,resources,andexpertisetothegovernanceprocess.Together,wecantacklethesignificant

workneededtoestablishrobust,effective,andgloballyconsistentframeworks.

weidentifiedkeyrecommendationsforwhat

differentactorscandotoadvanceAIgovernance.Toderivetheserecommendations,weanalyzed

keyinsightsfromeachsessionandcategorized

thembasedontheactorswhocouldmostbenefit.

Thisapproachsynthesizedwhatwasvoiced

duringthediscussionsandtransformedthose

insightsintotangible,actionable

recommendations.Thefollowingoverviewaimstoinspiredeeperconversationsonpractical,

tangibleactionseachcantaketobolster

accountabilityandupholdsafetyandfundamentalrights,includingtheruleoflaw.

EstablishEnforceableOversight

Movebeyondrelianceonvoluntaryindustry

commitmentstomandatoryrequirementsby

creatingbindinglegalmandates.Adopt

comprehensiveoversightmechanismsandclearredlinesforAIdevelopment—drawingontheEUAIActandFrontierAISafetyCommitments—to

preventrecklessdeploymentandprotectpublicinterest.

RequireRigorousTesting,IndependentEvaluations,andThird-PartyAudits

MandatefrequentandstandardizedindependentassessmentsofAIalignedwithinternational

5

standards.

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

INTERGOVERNMENTALORGANIZATIONS

DevelopGlobalStandards

Collaboratewiththe

InternationalNetworkofAI

SafetyInstitutes

andtheOECDtoestablishandcontinuouslyupdateinteroperablestandardsforAIgovernance,leveragingthemomentumofthe

AIActionSummitconsultations

,andensuring

thesestandardsareadopted,orattheveryleast,recognizedacrossmultiplejurisdictions.

CreateBindingInternationalFrameworks

Supportagreementsthatrequirecompaniesandgovernmentstoadoptspecificandclearoversightprotocolsandmeasurablecompliance

requirements,includingwithclearincentivesandredlinesthatalignwiththeEUAIAct.These

frameworksshouldalsoenableinteroperabilitywithevolvingglobalregulatoryinitiatives

worldwide.

FosterGlobalCooperation

Institutionalizejointoversightmechanisms—

especiallythroughapermanentstructurefortheInternationalNetworkofAISafetyInstitutes—to

addresscross-borderrisks,supportaholistic

incidentreportingplatforminpartnershipwiththeOECD,andscalepoliciesforaccidentpreventionglobally.

mechanismsandincidentreportingrequirementsacrossallstagesofAIdevelopmentand

deployment.

CommittoLifecycleAccountability

TakefullresponsibilityforAIthroughouttheir

lifecycle,includingproactivelymonitoringtheir

impactspost-deployment.Addressunintended

consequencesswiftlybyimplementingcorrectivemeasuresandprovidingtransparentupdatesto

stakeholders.

CIVILSOCIETY

EnableInclusiveParticipationinAIGovernance

Buildcoalitionsacrosscivilsocietyorganizationstoensureunderrepresentedvoicesareactivelyincludedindecision-makingforums.Foster

equitablerepresentationandcreateavenuesformeaningfulcontributionstogovernance

discussions.

MonitorandAssessAIImpacts

Establishindependent,well-fundedinitiativestosystematicallytrack,analyze,andpubliclyreportonthesocietalimpactsofAI.Usethese

assessmentstoholdactors(primarily

governmentsandAIcompanies)accountableforharms,biases,andunintendedconsequences,

whileadvocatingforlegallyenforceablemechanisms.

INDUSTRY

ImplementTransparencyMeasures

Regularlypublishclear,detailedreportsdetailingcompliancewithethicalguidelinesand

regulations,adherencetosafetystandards,

outcomesofriskassessmentsandmodel

evaluationsandtesting,andmitigationstrategies.Thesereportsshouldalignwithinternationally

recognizeddisclosureandaccountabilitystandards.

EnsureResponsibleandTrustworthyDevelopmentofAI

Developandimplementcomprehensivesafetyprotocolsthatincludeclearethicaluse

restrictions,robuststakeholderoversight

DevelopAdaptiveGlobalBenchmarks

Leadeffortstocreateandupdaterigorous,

globallyrelevantevaluationframeworksforAI.

Thesebenchmarksshouldstandardizemetricsforperformance,safety,andsocietalimpact,ensuringconsistencyandaccountabilitywhileadaptingtorapidtechnologicaladvancements.

INDIVIDUALS

AdvocateforTransparencyandRights

DemandgreatertransparencyfromAIdevelopersandpolicymakers,ensuringAIalignswithsocietalvaluesandpublicinterestthroughactivecivic

education.

6

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

PushforEnforceableGovernance

Callforstrongeraccountabilitymeasuresfrom

governmentsandcorporations,ensuringAIservesthepublicinterest.

ParticipateinPublicConsultations

ContributetodiscussionsandpolicymakingonAIgovernancebyengaginginconsultationsand

forums.Diverseperspectivesarecriticalforrobustaccountabilitymechanisms.

SupportEthicalAIDevelopment

Activelysupportdevelopersandorganizationsthatprioritizetransparencyandaccountability.

Conversely,rejectcompaniesorproductsthatfailtomeetethicalstandardsoraddresstheharm

causedbytheirtechnologies—forexample,bychoosingnottosupportorpurchasefromthem.

7

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

TheSixthEditionInNumbers

TheAthensRoundtableconvenedadiverseaudience,allowingforvariedandnuancedperspectivesonglobalAIgovernance.Belowarekeystatisticshighlightingtheevent’sscopeandimpact.

ATTENDANCE

PA

108

COUNTRIESREPRESENTED

PA

Numberofparticipants

AttendanceBreakdownIIn-personVs.onlineparticipants

144

TOTAL

INTERVENTIONS

SPOKENINTERVENTIONS

>

CI

NumberofInterventions

InterventionsBreakdownIPlannedVs.spontaneous;civilsociety

8

SECTOR

REGIONS/COUNTRIES

9

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

GENDER

POLLINGOFONLINEPARTICIPANTS

QUESTION|HowconfidentareyouthatAIwillbegovernedproperlytoensureitissafeandtrustworthy?

TimeofDay

changeinpercentageofAttendeesMarkingTheirconfidence

asLow(1-6/10)ThroughoutTheDay

10

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

11

AgendaOverview

Forthisedition,TheAthensRoundtablefocused

onenhancinginternationalcoordinationfor

accountabilityinAIgovernanceandservedasanofficialeventleadingto

France’sAIActionSummit

inFebruary2025.Theeventbeganwithopeningremarksemphasizingtheneedforrobust

mechanismstosafeguardtheruleoflaw,

democraticvalues,andequitableglobal

innovation.Theroundtablediscussionsfollowedadynamicformat,combiningshortinterventions

fromparticipantsandinteractiveexchanges.

Onlineaudienceengagedinrealtimethroughpollsandchat.

Session1mappedtheemerginglandscapeofAIgovernance,withrepresentativesfromkey

institutionssuchastheOECD,UnitedNations,

UNESCO,AISafetyInstitutes,andtheEuropeanAIOfficeshowcasingtheirrespective

accountabilityinitiatives.

Session2examinedcoregapsinestablishing

accountableAIgovernance,includingthedriveforstrongerglobalstandards,clarityonliability

frameworks.StrategiesformanagingAIrace

dynamics,andensuringdiverseperspectives—especiallyfromtheGlobalSouth—wereheard.

Thesessionalsoemphasizedtheneedforpracticaltoolstoclosethesegaps,suchasadaptablepolicyframeworksandinclusivedecision-makingprocesses.

Session3exploredwaystobuildstronger

accountabilityacrossAIactors,delvinginto

incentivesforresponsibledevelopment,

interoperableevaluations,transparentinformationflows,andeffectiveincidentreporting.

Session4announcedthe

findingsoftheAIAction

SummitConsultation

,reflectinginputfromover

10,000citizensand200civilsocietyorganizationson‘“constructivevigilance”topreventAIharms

andnecessaryregulatorymeasures.Thesessionhighlightedtheimportanceofleveragingthese

findingstoensurethatcivilsocietyvoicesdirectlyshapeglobalgovernanceeffortsandinfluence

policyprioritiesaheadoftheSummit.Italso

addressedthecentralroleofcivilsociety

organizationsinshapingglobalAIgovernanceandproposedforgingadedicatedcoalitiontounifytheirvoices.

Participantsclosedbydiscussingnextstepsto

maintainmomentumleadingintotheAIAction

Summit.CallstoembedaccountabilitythroughouttheAIlifecycle,protecthumanrights,and

reinforcedemocraticprinciplesunderscoredtheentiredialogue.Buildingoninsightssharedanddrawingfromtheconsultationresults,thisreportdetailstheconcreteobjectivesanddeliverablesproposedduringtheRoundtable.

MoreinformationontheSixthEdition’sagenda,includingafulllistofin-personparticipants,is

here

.

BuildingeffectivegovernanceforAIisnotanendpointbutan

ongoingendeavor.Thestakesarehigh:withouteffective

accountabilitymechanisms,AIwillexacerbateinequality,erodehumanrights,anddestabilizeourdemocraticinstitutions.

12

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

SessionTakeaways

OpeningRemarks

SPEAKERS

MinisterDimitrisPapastergiou

Minister,HellenicMinistryofDigitalGovernance

ChloéGoupille

SecretaryGeneral,AIActionSummit

UlrikVestergaardKnudsen

DeputySecretaryGeneral,OECD

NickMo?s

ExecutiveDirector,TheFutureSociety

Theopeningremarkssetthestagefortheday’sdiscussionsbyhighlightingthecriticalneedfor

accountabilityinAIgovernance.Speakersunderscoredtheurgencyofaddressingglobalchallengesandemphasizedthatfacilitatingcollaborationandinteroperabilitybetweencountries,companies,andthe

ecosystemasawholeisessentialforAIgovernancetosucceed.Theremarksestablishedacallfor

broadersocietalparticipation,particularlyfromglobalmajoritycountriesandcivilsociety,toensureAIgovernancemechanismsaremorereflectiveofdiverseperspectivesworldwide.

13

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

KEYINSIGHTS

Human-CentricAIDevelopment

MinisterDimitrisPapastergioucalledfora

humanisticandenvironmentallysustainable

approachtoAIdevelopment.Thisapproach

emphasizesthreethings:protectionofminors,

equitablebenefitdistribution,andsafeguarding

fundamentalrights.TheMinisteradvocatedforthepreservationofculturaldiversitythroughlocalAI,particularlyforlesserspokenlanguageslike

Greek,andproposedthecreationofaEuropeanopen-sourcelargelanguagemodel.Thisinitiativeseekstoreducedependencyonthird-party

systemsandpromoteregionalinnovationwhiledemocratizingAIaccessandownership.

RisksofRegulatoryFragmentation

UlrikVestergaardKnudsendrewattentiontothedangersofregulatoryfragmentation,emphasizingtheresultinginefficiencies,legalcomplexities,andriskstotheglobaleconomy.Headvocatedfor

interoperabilityandconsensus-buildingacrossgovernments,internationalorganizations,civil

society,andtheprivatesector.Tocountertheserisks,Knudsencalledforthedevelopmentof

cohesive,globalAIgovernanceframeworksthat

alignwithdemocraticprinciplesandsharedvalues,promotingeconomicandregulatorystability.

PavingtheWayforInclusiveandAction-OrientedAIGovernance

ChloéGoupilleframedtheAIActionSummitasatransformativeopportunitytodriveconsensusandimplementactionablepracticalpolicies.She

stressedtheimportanceofcreatinga

coordinated,democraticgovernanceframeworkthatprioritizespublicinterestandsecuritywhileamplifyingdiverseperspectives.

UrgencyinDevelopingAccountabilityMechanisms

NickMo?semphasizedtheneedforpragmatic

accountabilitymechanismstorestorepublictrustandensureresponsibleAIdevelopment.Mo?s

highlightedtheimportanceofinstitutional

innovationtoaddressgapsingovernanceand

urgedparticipantstoleavetheroundtablewith

clearandactionablesolutionsfortheir

organizations.Mo?sunderscoredtheimportanceofbalancinginnovationwithsocietalsafeguards,cautioningagainstmoving“toofastandbreakingtoomanythings.”

SessionTake

aways14

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

KEYQUOTES

Let'snotpretendotherwise:theelephantintheroomisthe

excessiveconcentrationofinformation,algorithms,traffic,andwealthinextremelyfewcenters.AImustbedemocratized,

openedupand[ableto]attractmoreowners,advocates,andsupporters.

MinisterDimitrisPapastergiou

Minister,HellenicMinistryofDigitalGovernance

AccountabilitymeansthatAIactorsshouldapplyasystematicriskmanagementapproachtoeachphaseoftheAIsystem

lifecycleandadoptresponsiblebusinessconducttoaddressrisksrelatedtoAI.

UlrikVestergaardKnudsen

DeputySecretaryGeneral,OECD

TheAIActionSummit,plannedinFebruary2025,isagreat

opportunity[for]awiderangeofstakeholders,entrepreneurs,andgovernmentofficialstotakedecisions,but,most

importantly,actions.

ChloéGoupille

SecretaryGeneral,AIActionSummit

Wewillhavetodesignnewinstitutionalsolutions,tokeep

industryandourselvesaccountable,andtopreventusfrommovingtoofastandbreakingtoomanythings.

NickMo?s

ExecutiveDirector,TheFutureSociety

Ses

sionTakeaways15

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

RELEVANTONLINEPOLLRESULTS

>

percentage(%)

pollResultsIWhichareasshouldweprioritizetostrengthen

GlobalAAccountability?

SessionTakeaways16

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

SESSION1

MappingtheLandscapeofAIGovernance

andAccountability

OPENINGINTERVENTIONSBY

CarmeArtigas

Co-ChairattheAIAdvisoryBoard,OfficeoftheUNSecretary-

General'sEnvoyonTechnology

CédricWachholz

ChiefofDigitalInnovationandTransformationSection,

UNESCOCommunicationandInformationSector

ElizabethKelly

Director,

U.S.AISI,NIST

AudreyPlonk

DeputyDirector,OECD

DirectorateforScience,

TechnologyandInnovation

LucillaSioli

Director,

EuropeanAIOffice

17

NUMBEROFSPEAKERS

DURINGOPENDISCUSSION

6

NUMBEROFCIVILSOCIETYINTERVENTIONS

Inthisinitialsession,representativesfromleadingglobalAIgovernanceinstitutions–theEuropeanAI

Office,theU.S.AISafetyInstitute,theOECD,UN,andUNESCO–mappedtheevolvingAIgovernance

landscape.SpeakersandtheaudienceoutlinedkeymeasurestostrengthenaccountabilityintheglobalAIecosystem,emphasizinginclusivity,transparency,andinteroperability.

SessionTakeaways17

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

KEYINSIGHTS

DebateAbout“OpenSource”AI

Therewassignificantdisagreementaboutthe

marginalriskofopensourceAI.JackClarkfromAnthropicraisedconcernsaboutmisuseofopenweights,citingcaseswhereimagegenerators

wereusedforchildexploitationaftersafety

featureswereremoved.Severalothersargued

thatopensourceincreasestransparencyand

accountability,drawingparallelstocybersecuritywhereopensystemshaveprovensecure.

BenjaminPrud'homme,Mila-QuebecArtificial

IntelligenceInstitute,suggestedmovingpasttheopen/closedbinarytodiscussspecific"model

sharingmodalities."

EthicsversusGovernance

CarmeArtigashighlightedthelimitationsofethicalAIguidelines,notingthattheyoftenlack

enforceableaccountabilitymechanisms,are

subjective,canshiftwithchangesinleadership,

andareculturallydependent,varyingbycontext.Shearguedthatgovernancemustcomplement

ethicalguidelinesbyestablishingconcrete

instrumentsandmechanismstoensurethat

companiesandgovernmentsactethicallyandareheldaccountable.LucillaSioliemphasizedthe

growingconcernsaroundtherisksAIposedto

fundamentalrightsandsafety.February2025,inherview,willmarkacriticalmilestoneinthis

effort,withapproximately1,000stakeholdersfromindustryandcivilsocietyactivelycontributingto

thedraftingoftheGeneral-PurposeAICodeofPractice.

TheLimitsofVoluntaryCommitments

Manyspeakersunderscoredtheinsufficiencyof

voluntarycommitmentsforeffectiveAI

governance.GaiaMarcusfromtheAdaLovelaceInstituteassertedthat“voluntarycommitmentsarenolongerdefensible”duetocompetingprivate

sectorincentives.PaulNemitzcalledrelianceon

voluntarymeasures“naive,”advocatinginstead

forenforceablerightsofactionforbothindividualsandcivilsociety—againstcompaniesand

enforcementbodiesalike.

LucillaSiolihighlightedtheEuropeanUnion’s

regulatoryapproachasanexampleof

enforceableframeworksinaction.By

implementingacomprehensiveframework,the

EUhastakensignificantstepstomitigateriskstofundamentalrightsandsafety.Thisservesasa

clearcontrasttothelimitationsofvoluntary

measures,reinforcingtheneedforstronger

regulatoryoversightinaddressingthechallengesposedbyAI.

Effectivegovernancemust

addressbothmarket

concentrationandthe

disparitiesinaccessand

capabilitiestoensureAIbenefitsaredistributedequitably.

ConcentrationofPower

Marketconcentrationwasidentifiedasamajor

challengeforAIgovernance.SebastianoToffalettioftheEuropeanDigitalSMEAlliancecomparedAIdevelopmenttotheopeninternet,where

“everybodyhastherighttocreatetheirown

website,”arguingthatthecentralizationofAI

developmentamongafewplayersundermines

accountability.WhilesomeemphasizedtheneedtodecentralizeAIdevelopment,othersargued

thatgovernancemechanismsmustaddresschallengesregardlessofmarketstructure.

CedricWachholzhighlightedanotherdimensionofthisissue:unequalaccesstoAItoolsand

training.WhilemanyareusingtoolslikeChatGPT,onlyasmallfractionhaveaccesstoproper

guidanceoreducation.Thislackofaccessrisksdeepeninginequalities,asthosewiththe

SessionTakeaways18

knowledgetoleverageAItoolsgainsignificantadvantages.EffectivegovernancemustaddressbothmarketconcentrationandthedisparitiesinaccessandcapabilitiestoensureAIbenefitsaredistributedequitably.

THEATHENSROUNDTABLE

ARTIFICIALINTELLlGENCEANDTHERULEOFLAW

KEYQUOTES

[T]hereneedstobeanimpartialthirdpartythatlooksatthe

potentialrisksormisusesofAI,whetherproprietaryoropen,orthisdiscussionwillbeincrediblyunproductiveandheated

forever,andwe'llneversolveanything.

JackClark

Co-FounderandHeadofPolicy,Anthropic

Voluntarycommitmentsarenolongerdefensible.Iwouldsaytherearetoomanyotherincentivesonprivatelybased

companiestojustexpectthemtorespondtovoluntarycommit