A charter on the use of artificial intelligence was created in Paris. Ten commandments for journalists

We, as rep­re­sen­ta­tives of the media and jour­nal­ism com­mu­ni­ty, acknowl­edge the trans­for­ma­tive impli­ca­tions of arti­fi­cial intel­li­gence (AI) for human­i­ty. We cham­pi­on glob­al coop­er­a­tion to ensure AI upholds human rights, peace and democ­ra­cy, and aligns with our shared aspi­ra­tions and val­ues. The his­to­ry of news and infor­ma­tion is inter­twined with tech­no­log­i­cal progress. AI, span­ning from basic automa­tion to ana­lyt­i­cal and cre­ative sys­tems, intro­duces a new cat­e­go­ry of tech­nolo­gies with an unpar­al­leled capac­i­ty to inter­sect with human thought, knowl­edge, and cre­ativ­i­ty. It rep­re­sents a con­sid­er­able shift in infor­ma­tion gath­er­ing, truth seek­ing, sto­ry­telling, and the dis­sem­i­na­tion of ideas. As such, it will pro­found­ly alter the tech­ni­cal, eco­nom­ic and social con­di­tions of jour­nal­ism and edi­to­r­i­al prac­tice.

AI sys­tems have the poten­tial, depend­ing on their design, gov­er­nance and appli­ca­tion, to rev­o­lu­tion­ize the glob­al infor­ma­tion land­scape. How­ev­er, they also present a struc­tur­al chal­lenge to the right to infor­ma­tion. The right to infor­ma­tion flows from the free­dom to seek, receive and access reli­able infor­ma­tion. It is root­ed in the inter­na­tion­al legal frame­work, includ­ing the Uni­ver­sal Dec­la­ra­tion of Human Rights, the Inter­na­tion­al Covenant on Civ­il and Polit­i­cal Rights, and the Inter­na­tion­al Part­ner­ship for Infor­ma­tion and Democ­ra­cy.

This right under­pins the fun­da­men­tal free­doms of opin­ion and expres­sion. The social role of jour­nal­ism and media out­lets —serv­ing as trust­wor­thy inter­me­di­aries for soci­ety and indi­vid­u­als— is a cor­ner­stone of democ­ra­cy and enhances the right to infor­ma­tion for all. AI sys­tems can great­ly assist media out­lets in ful­fill­ing this role, but only if they are used trans­par­ent­ly, fair­ly and respon­si­bly in an edi­to­r­i­al envi­ron­ment that staunch­ly upholds jour­nal­is­tic ethics. In affirm­ing these prin­ci­ples, we uphold the right to infor­ma­tion, cham­pi­on inde­pen­dent jour­nal­ism, and com­mit to trust­wor­thy news and media out­lets in the era of AI.

1. JOURNALISM ETHICS GUIDE THE WAY MEDIA OUTLETS AND JOURNALISTS USE TECHNOLOGY

Media out­lets and jour­nal­ists use tech­nolo­gies that enhance their capac­i­ty to ful­fill their pri­ma­ry mis­sion: ensur­ing everyone’s right to qual­i­ty, trust­wor­thy infor­ma­tion. The pur­suit and achieve­ment of this goal should dri­ve their choic­es regard­ing tech­no­log­i­cal tools. The use and devel­op­ment of AI sys­tems in jour­nal­ism must uphold the core val­ues of jour­nal­is­tic ethics, includ­ing truth­ful­ness and accu­ra­cy, fair­ness, impar­tial­i­ty, inde­pen­dence, non-harm, non-dis­crim­i­na­tion, account­abil­i­ty, respect for pri­va­cy and for the con­fi­den­tial­i­ty of sources.

2. MEDIA OUTLETS PRIORITIZE HUMAN AGENCY.

Human deci­sion-mak­ing must remain cen­tral to both longterm strate­gies and dai­ly edi­to­r­i­al choic­es. The use of AI sys­tems should be a delib­er­ate and wellinformed deci­sion made by humans. Edi­to­r­i­al teams must clear­ly define the goals, scope, and usage con­di­tions for each AI sys­tem. They must ensure a cross-sec­tion­al and con­tin­u­ous over­sight of the impacts of deployed AI sys­tems, ensure their strict com­pli­ance with their usage frame­work, and retain the abil­i­ty to deac­ti­vate them at any time.

3. AI SYSTEMS USED IN JOURNALISM UNDERGO PRIOR, INDEPENDENT EVALUATION

The AI sys­tems used by the media and jour­nal­ists should under­go an inde­pen­dent, com­pre­hen­sive, and thor­ough eval­u­a­tion involv­ing jour­nal­ism sup­port groups. This eval­u­a­tion must robust­ly demon­strate adher­ence to the core val­ues of jour­nal­is­tic ethics. These sys­tems must respect pri­va­cy, intel­lec­tu­al prop­er­ty and data pro­tec­tion laws. A clear account­abil­i­ty frame­work is estab­lished for any fail­ure to meet these require­ments. Sys­tems that oper­ate pre­dictably and can be sim­ply explained are pre­ferred.

4. MEDIA OUTLETS ARE ALWAYS ACCOUNTABLE FOR THE CONTENT THEY PUBLISH

Media out­lets assume edi­to­r­i­al respon­si­bil­i­ty, includ­ing in their use of AI in gath­er­ing, pro­cess­ing, or dis­sem­i­nat­ing infor­ma­tion. They are liable and account­able for every piece of con­tent they pub­lish. Respon­si­bil­i­ties tied to the use of AI sys­tems should be antic­i­pat­ed, out­lined, and assigned to humans to ensure adher­ence to jour­nal­is­tic ethics and edi­to­r­i­al guide­lines.

5. MEDIA OUTLETS MAINTAIN TRANSPARENCY IN THEIR USE OF AI SYSTEMS

 Any use of AI that has a sig­nif­i­cant impact on the pro­duc­tion or dis­tri­b­u­tion of jour­nal­is­tic con­tent should be clear­ly dis­closed and com­mu­ni­cat­ed to every­one receiv­ing infor­ma­tion along­side the rel­e­vant con­tent. Media out­lets should main­tain a pub­lic record of the AI sys­tems they use and have used, detail­ing their pur­pos­es, scopes, and con­di­tions of use.

6. MEDIA OUTLETS ENSURE CONTENT ORIGIN AND TRACEABILITY

Media out­lets should, when­ev­er pos­si­ble, use state-of-theart tools that guar­an­tee the authen­tic­i­ty and prove­nance of pub­lished con­tent, pro­vid­ing reli­able details about its ori­gin and any sub­se­quent changes it may have under­gone. Any con­tent not meet­ing these authen­tic­i­ty stan­dards should be regard­ed as poten­tial­ly mis­lead­ing and should under­go thor­ough ver­i­fi­ca­tion.

7. JOURNALISM DRAWS A CLEAR LINE BETWEEN AUTHENTIC AND SYNTHETIC CONTENT

Jour­nal­ists and media out­lets strive to ensure a clear and reli­able dis­tinc­tion between con­tent derived from the phys­i­cal cap­ture of the real world (such as pho­tographs, and audio and video record­ings) and that which is cre­at­ed or sig­nif­i­cant­ly altered using AI sys­tems. They should favor the use of authen­tic footage and record­ings to depict actu­al events. Media out­lets must avoid mis­lead­ing the pub­lic in their use of AI tech­nolo­gies. In par­tic­u­lar, they should refrain from cre­at­ing or using AI-gen­er­at­ed con­tent mim­ic­k­ing real-world cap­tures and record­ings or real­is­ti­cal­ly imper­son­at­ing actu­al indi­vid­u­als.

8. AI-DRIVEN CONTENT PERSONALIZATION AND RECOMMENDATION UPHOLDS DIVERSITY AND THE INTEGRITY OF INFORMATION

In media out­lets, the design and use of AI sys­tems for auto­mat­ic con­tent per­son­al­iza­tion and rec­om­men­da­tion should be guid­ed by jour­nal­is­tic ethics. Such sys­tems should respect infor­ma­tion integri­ty and pro­mote a shared under­stand­ing of rel­e­vant facts and view­points. They should high­light diverse and nuanced per­spec­tives on var­i­ous top­ics, fos­ter­ing open-mind­ed­ness and demo­c­ra­t­ic dia­logue. The use of such sys­tems must be trans­par­ent, and users should when­ev­er pos­si­ble be giv­en the option to dis­able them to ensure unfil­tered access to edi­to­r­i­al con­tent.

9. JOURNALISTS, MEDIA OUTLETS AND JOURNALISM SUPPORT GROUPS ENGAGE IN THE GOVERNANCE OF AI

As essen­tial guardians of the right to infor­ma­tion, jour­nal­ists, media out­lets and jour­nal­ism sup­port groups should play an active role in the gov­er­nance of AI sys­tems. They should be includ­ed in any glob­al or inter­na­tion­al insti­tu­tion­al over­sight of AI gov­er­nance and reg­u­la­tion. They should ensure that AI gov­er­nance respects demo­c­ra­t­ic val­ues, and that diver­si­ty of peo­ple and cul­tures is reflect­ed in the devel­op­ment of AI. They must remain at the fore­front of knowl­edge in the field of AI. They are com­mit­ted to exam­in­ing and report­ing on the impacts of AI with accu­ra­cy, nuance, and a crit­i­cal mind.

10. JOURNALISM UPHOLDS ITS ETHICAL AND ECONOMIC FOUNDATION IN ENGAGEMENTS WITH AI ORGANIZATIONS

Access to jour­nal­is­tic con­tent by AI sys­tems should be gov­erned by for­mal agree­ments that ensure the sus­tain­abil­i­ty of jour­nal­ism and uphold the long-term shared inter­ests of the media and jour­nal­ists. AI sys­tem own­ers must cred­it sources, respect intel­lec­tu­al prop­er­ty rights, and pro­vide just com­pen­sa­tion to rights holders.This com­pen­sa­tion must be passed on to jour­nal­ists through fair remu­ner­a­tion. AI sys­tem own­ers are also required to main­tain a trans­par­ent and detailed record of the jour­nal­is­tic con­tent uti­lized to train and feed their sys­tems. Rights hold­ers must make the repur­pos­ing of their con­tent by AI sys­tems con­di­tion­al on respect for the integri­ty of the infor­ma­tion and the fun­da­men­tal prin­ci­ples of jour­nal­is­tic ethics. They col­lec­tive­ly call for AI sys­tems to be designed and used in such a way as to guar­an­tee high-qual­i­ty, trust­wor­thy and plu­ral­is­tic infor­ma­tion.

Read more:

Bialystok: Keeping touch burning for imprisoned journalists

London calling. Anti-censorship activists took part in Solidarity Marathon for Belarusian Journalists

The Internet Governance Forum aims to legitimize suppression of free speech