As Trinidad and To­ba­go pre­pares for its up­com­ing gen­er­al elec­tion, a lead­ing tech­nol­o­gy ex­pert is rais­ing red flags about the grow­ing role of ar­ti­fi­cial in­tel­li­gence in po­lit­i­cal cam­paign­ing.

Dig­i­tal an­thro­pol­o­gist Daren Dho­ray is warn­ing that the tech­nol­o­gy has the po­ten­tial to ma­nip­u­late pub­lic opin­ion and un­der­mine de­mo­c­ra­t­ic process­es.

Dho­ray fur­ther cau­tioned that AI-dri­ven tools, such as Chat­G­PT, could pose sig­nif­i­cant threats to elec­tion in­tegri­ty if left un­reg­u­lat­ed.

“Whether it is cur­rent­ly play­ing a role—I wouldn’t want to say no—but I’m al­most cer­tain that AI is in­flu­enc­ing what the gen­er­al pub­lic thinks about po­lit­i­cal par­ty A or par­ty B, es­pe­cial­ly since all po­lit­i­cal par­ties are now uti­liz­ing so­cial me­dia for much of their cam­paign mes­sag­ing. It makes it dif­fi­cult to dis­cern what is re­al and what is fake.”

Dho­ray’s com­ments come amid al­le­ga­tions from Op­po­si­tion Leader Kam­la Per­sad-Bisses­sar that mem­bers of the rul­ing Peo­ple’s Na­tion­al Move­ment (PNM) are us­ing AI to spread fake news as part of their cam­paign strat­e­gy.

Re­cent­ly, an au­dio clip sur­faced pur­port­ing to fea­ture a con­ver­sa­tion be­tween Per­sad-Bisses­sar and To­ba­go House of As­sem­bly (THA) Chief Sec­re­tary Far­ley Au­gus­tine, al­leged­ly dis­cussing col­lab­o­ra­tion ahead of the elec­tion. Per­sad-Bisses­sar has since la­belled the clip as fake.

Prime Min­is­ter Stu­art Young has pushed back, sug­gest­ing there is ev­i­dence that the Unit­ed Na­tion­al Con­gress (UNC) has al­so been us­ing AI to cre­ate fake so­cial me­dia pro­files and gen­er­ate mis­lead­ing com­ments on­line.

Dho­ray said he would not be sur­prised if ma­jor po­lit­i­cal par­ties have al­ready in­cor­po­rat­ed AI in­to their cam­paign strate­gies—and high­light­ed the risks that come with it.

“It all comes back to the laws we have re­gard­ing how da­ta can and can­not be used. While we do have some da­ta pro­tec­tion and pri­va­cy leg­is­la­tion, it might be a stretch to use those laws to charge some­one specif­i­cal­ly for mis­in­for­ma­tion. Eth­i­cal­ly, there’s al­so the ques­tion: is it moral­ly right to use some­one’s im­age or voice with­out con­sent? In cas­es like that, the law could be called up­on un­der defama­tion of char­ac­ter.”

Dho­ray urged the pub­lic to re­main ex­tra vig­i­lant, not­ing that it’s be­com­ing in­creas­ing­ly dif­fi­cult to iden­ti­fy AI-gen­er­at­ed con­tent.

“It’s quite easy now. For those who aren’t fa­mil­iar with the tech­nol­o­gy, it’s im­por­tant to un­der­stand that a What­sApp video be­ing shared might not ac­tu­al­ly fea­ture the per­son it ap­pears to. It could very well be a fake video or im­age. This makes it much hard­er for those who are less tech-savvy.”

He main­tained that while AI presents in­no­v­a­tive op­por­tu­ni­ties for im­prov­ing po­lit­i­cal cam­paigns and vot­er en­gage­ment, it al­so rais­es se­ri­ous con­cerns re­lat­ed to mis­in­for­ma­tion, cy­ber­se­cu­ri­ty, and ethics.

“AI-con­trolled chat­bots can flood so­cial me­dia with co­or­di­nat­ed mes­sages that am­pli­fy dis­in­for­ma­tion or sup­press op­pos­ing views. Phish­ing is al­ready a ma­jor is­sue, but now AI makes it easy to au­to­mate phish­ing at­tacks that mim­ic of­fi­cial elec­tion com­mu­ni­ca­tions. This can trick peo­ple in­to shar­ing sen­si­tive da­ta, or even lead to the cre­ation of fraud­u­lent fundrais­ing web­sites that spoof le­git­i­mate cam­paign pages—steal­ing do­na­tions from un­sus­pect­ing sup­port­ers.”