Raphael John-Lall

While there are op­por­tu­ni­ties giv­en the break­neck speed with which Ar­ti­fi­cial In­tel­li­gence (AI) is evolv­ing, com­pa­nies and busi­ness­es must safe­guard them­selves from po­ten­tial threats that are emerg­ing and should “get their house in or­der.”

This is the ad­vice of man­ag­ing di­rec­tor of Pina­ka Con­sult­ing Ltd, Shi­va Bisses­sar who spoke at a we­bi­nar en­ti­tled “Im­pli­ca­tions of AI and Cy­ber­se­cu­ri­ty” host­ed by the Caribbean Cor­po­rate Gov­er­nance In­sti­tute last Fri­day.

Bisses­sar start­ed the we­bi­nar by show­ing a newsclip from Cana­da’s me­dia house, CN­BC, on the rapid and shock­ing ad­vances by Chi­na’s Ar­ti­fi­cial In­tel­li­gence com­pa­ny, DeepSeek, and the dis­com­fort it has caused in Chi­na’s main glob­al com­peti­tor, the Unit­ed States.

He said DeepSeek can be used by good ac­tors but al­so ma­li­cious ac­tors.

Bisses­sar said what is hap­pen­ing with AI now is very in­ter­est­ing, it is im­por­tant to note that AI start­ed years ago back in the 1950s and there have been dis­tinct pe­ri­ods of in­ac­tiv­i­ty and spurts of in­no­va­tion.

“Where we are now is in the hock­ey stick phase where there is ex­po­nen­tial growth, so some­thing like this tak­ing place is to be ex­pect­ed and it al­so brings up a lot of fears that peo­ple have with re­spect to ar­ti­fi­cial gen­er­al in­tel­li­gence and ar­ti­fi­cial su­per in­tel­li­gence. Pre­vi­ous­ly, we have had sim­ple chat­bots, we have had Siri and Alexa, in­ter­ac­tive ways to is­sue sim­ple in­struc­tions, re­trieve sim­ple in­struc­tions, But now what is all the rage with AI is the gen­er­a­tive as­pect, that abil­i­ty to gen­er­ate con­tent, texts, im­ages, video, sounds. It re­al­ly start­ed in 2022 with the re­lease of Chat­G­PT.”

He re­count­ed that in 2024, he par­tic­i­pat­ed in a con­fer­ence in Gene­va, Switzer­land and was “com­plete­ly floored” by the num­ber of ap­pli­ca­tions that are be­ing de­vel­oped for AI.

“This has been tak­ing place qui­et­ly in the back­ground. The things that were in the pub­lic were the gen­er­a­tive as­pect with Open AI and Chat­G­PT in 2022. Be­yond what is pop­u­lar in the news, AI con­tin­ues to de­vel­op and flour­ish in the back­ground. From that con­fer­ence, a cou­ple of my key take­aways would have been things like the med­ical ap­pli­ca­tions, study­ing things like mul­ti-mo­bil­i­ty. An­oth­er take­away was deep­fake which cov­ers AI and cy­ber­se­cu­ri­ty. There is a con­stant con­flict be­tween the white hats or good guys and the black hats and the ma­li­cious in­tent guys.

He al­so spoke about some of the “fas­ci­nat­ing” de­vel­op­ments that are tak­ing place.

“It is get­ting pret­ty fas­ci­nat­ing what is pos­si­ble. We have now, some­thing akin to a fil­ter that could be ap­plied live over an im­age on so­cial me­dia and you would ap­pear as a dif­fer­ent per­son. We re­al­ly have some fear­ful ar­eas be­ing de­vel­oped. In Asia, an em­ploy­ee au­tho­rised a wire trans­fer, based on a video con­ver­sa­tion he had with one of his high­er ups but it turns out it was a deep­fake that he was con­vers­ing with. So, mov­ing from video to live deep fakes, that is sig­nif­i­cant. How do we de­tect and go up against that?”

Dur­ing the ques­tion-and-an­swer seg­ment, Bisses­sar was asked about that em­ploy­ee who was de­ceived af­ter he thought he was in­ter­act­ing with a com­pa­ny man­ag­er and what can com­pa­nies do to put sys­tems in place to pre­vent this from hap­pen­ing, Bisses­sar said busi­ness­es must have the prop­er sys­tems in place to pro­tect them­selves.

“It is still ear­ly stages with re­spect to it hap­pen­ing live on the spot but there have been demon­stra­tions of it. A few months ago, some­one was able to de­mo how it is pos­si­ble. They did a face swap with a cou­ple pop­u­lar celebri­ties. Every or­gan­i­sa­tion is ex­pect­ed to have poli­cies and pro­ce­dures as to how you man­age au­then­ti­ca­tion of par­ties, ver­i­fy you are talk­ing to some­one and en­sure it is who you are talk­ing to. That dy­nam­ic is chang­ing, this is play­ing out in front of us. So, you have not paid at­ten­tion to cy­ber­se­cu­ri­ty in the past, which I know takes place lo­cal­ly and you see it. Peo­ple do not have chief in­for­ma­tion se­cu­ri­ty of­fi­cers. With­in the Caribbean, you will find them in fi­nan­cial in­sti­tu­tions but pre­dom­i­nant­ly with­in or­gan­i­sa­tions in T&T, I would say there is a lack of that.”

He spoke about more neg­a­tive ef­fects of the evo­lu­tion of AI that will im­pact the Caribbean re­gion and T&T specif­i­cal­ly.

“I have had an ex­pe­ri­ence with a fi­nan­cial sex­tor­tion scam, which re­sult­ed un­for­tu­nate­ly in a teenag­er’s death. So, these scams are re­al and they could on­ly be­come hy­per­charged with un­fet­tered de­vel­op­ment of AI. This brings us to the points of ethics and prop­er gov­er­nance and this is why it is ap­pro­pri­ate at the cor­po­rate gov­er­nance lev­el.”

De­vel­op­ing strate­gies

Cy­ber­se­cu­ri­ty con­sul­tant Pa­tri­cia Rowe­Seale, who is based in Bar­ba­dos, said busi­ness are more vul­ner­a­ble be­cause of the threats cre­at­ed by the ad­vances of AI.

“You have the avail­abil­i­ty of a num­ber of AI mod­els like fraudG­PT, WolfG­PT and there are a cou­ple mod­els that were specif­i­cal­ly de­signed for ne­far­i­ous ac­tiv­i­ty. From a cy­ber­se­cu­ri­ty and gov­er­nance per­spec­tive, it has made it a whole lot eas­i­er for at­tacks to oc­cur. You see the neg­a­tive side of AI from a cy­ber­se­cu­ri­ty per­spec­tive. What can we do and what is the re­spon­si­bil­i­ty of board mem­bers in terms of giv­ing that com­fort and as­sur­ance that the or­gan­i­sa­tion is aware and not on­ly be­ing aware but be af­fec­tive­ly man­ag­ing.”

She urged com­pa­nies to adopt strate­gies to counter the threats.

“One of the things that the board has to be very cog­nisant of is what is the cy­ber­se­cu­ri­ty strat­e­gy? Is this strat­e­gy com­pre­hen­sive? Is it be­ing relooked at, things that are hap­pen­ing in the en­vi­ron­ment and in the in­dus­try. How are you op­er­a­tional­is­ing this strat­e­gy? Do you have the right skill sets? Do you have suf­fi­cient fi­nances? We know that ran­somware has be­come so­phis­ti­cat­ed and hard to de­tect by the hu­man eye. So, you need to have some­thing with­in your email sys­tem, with­in your da­ta sys­tem to counter that. So, the ques­tions that the Board will be ask­ing is what are you us­ing to counter what is com­ing at us from an AI per­spec­tive.”

She al­so spoke about eth­i­cal and le­gal is­sues in­volved.

“You must be able to re­port to the board that you are aware of these eth­i­cal and le­gal is­sues and this is how se­nior man­age­ment is man­ag­ing and gov­ern­ing this space.”

Founder and CEO of One and One Ed­u­ca­tion­al Ser­vices Ltd. Ri­car­do Allen, based in Ja­maica, said that be­cause of how quick­ly the tech­nol­o­gy is mov­ing, com­pa­nies are hav­ing a hard time keep­ing up.

“AI has al­ways been some­thing that has been dis­cussed but it has al­ways felt pie-in-the-sky, some­thing that we want to achieve but it will not be any­time soon. In No­vem­ber, 2022 we saw the re­lease of Chat­G­PT and it took every­body by sur­prise. Since then, there has been so much rapid de­vel­op­ment in the AI space. You have dif­fer­ent per­sons com­ing up with their own mod­els and it is mov­ing so fast. For or­gan­i­sa­tions, how do they keep up and keep up will re­main­ing se­cure? Al­so how do you not get left be­hind sim­ply be fear­ing it.”

He spoke about “En­ter­prise AI” and why it is im­por­tant and warned busi­ness­es about the risks of us­ing mod­els like DeepSeek.

En­ter­prise AI is the in­te­gra­tion of ad­vanced AI-en­abled tech­nolo­gies and tech­niques with­in large com­pa­nies to en­hance var­i­ous busi­ness func­tions.

“As com­pa­nies, how do we en­sure that giv­en the way it kind of works, how do we en­sure that we pro­tect our­selves and or­gan­i­sa­tions? In the case of DeepSeek, a lot of per­sons have been ask­ing is it safe. In En­ter­prise AI it is about host­ing your mod­els lo­cal­ly or in the cloud and lever­ag­ing your ex­ist­ing da­ta sources to en­sure you get mean­ing­ful re­spons­es.”