Efforts Underway To Establish New AI Legal Framework: Pavan Duggal

Artificial intelligence (AI) has brought with itself a plethora of distinct issues. Its use case may be vast but it is also accompanied by several important aspects. The United Nations has come up with the world's first global artificial intelligence resolution calling for all member nations to come up with appropriate strategy to promote AI.

Speaking at BW Businessworld Gen AI Summit 2024, Dr Pavan Duggal, Advocate, Supreme Court of India; Chief Executive, Artificial Intelligence Law Hub said, "AI is (also) being talked about in the Indian parliament. There were massive discussions on deepfakes and voice cloning. We only have secondary legislation on AI. The work has begun on building a new body of legal jurisprudence on AI."

Duggal also cited the case of the Italian privacy watchdog, which had earlier refused to allow ChatGPT in the country as it was found to be breaching their law. 

"There were some changes and then ChatGPT was back in Italy. In India, we already have an AI in newscaster and also in teaching. It's not just about being a toy for adults. It's about being a cogent agent for protecting cyber security," Duggal added.

Further, Duggal said, "We are seeing a phenomenon where we are data objects. The IPR can be owned by the generator. A large number of legal nuances will come to fore which will have to be dealt with."

"The output of Generative AI will be legal. The authorship can be argued and also disputed. That will be a question of fact and not law. Litigation is not waiting for law to evolve. Litigation will help law evolve. Cases in US pertaining to Open AI are now being sued by left right and centre for damages. The Japanese authorities have come to the conclusion that using an AI model to train on a copyrighted work, then it is a fair use. That's two arguments are on two ends of spectrum." - Dr Pavan Duggal, Advocate, Supreme Court of India; Chief Executive, Artificial Intelligence Law Hub

Legality Of AI

"As humans we enter into contract with someone with an ability to contract. But in case of contract with AI, the question which needs to be asked as to if AI has the ability to contract. Whether it is a contract between the user and algorithm or user and company," Duggal said. 

Duggal highlighted the fundamental question, which was the ‘legality of AI’.

"Can it be granted legal status to AI? If yes, we will have to stipulate rights and duties. Almost all countries have taken the stand that AI cannot be an inventor, hence we cannot grant patents to it. However, South Africa has taken a contrarian position. Will the company providing AI ensure that it will not have bias or discriminatory. The 2024 will have massive use of deepfakes. The Pakistan elections there was extensive use of deepfakes. This is a big challenge for the Election Commission," Duggal said.

Highlighting the right to privacy, Duggal said that the privacy of Indians is also beginning to be impacted. 

"AI is an animal and we cannot leave it alone. There has to be minimal regulation surrounding it. We cannot allow AI to be wild," said Duggal.

Duggal said that the emergence of AI crime is beginning to come up as a new category within cyber crime.

Duggal further said that, "Cyber criminals do not want to dirty their hands and they can get the crime committed through an algorithm. So, a couple lost 15 lakh rupees. Their daughter was allegedly kidnapped and a ransom of Rs. 15 lakh was demanded within 15 minutes. The daughter was crying asking the father to save her. At the 14th minutes, the parents decided to pay. The school principal tells the father that his daughter has been in school."

"Another important issue is the relationship between AI and cyber security. We should not presume AI to be cyber secure because ultimately it is at best a computer resource. Hacking AI algorithm becomes a good option for cyber criminals," Duggal added.

Globally, countries are still in no man's land. Different countries are legislating national laws for AI. 

Duggal emphasised that there could not be any one size fits all approach. 

"There will have to be sector specific regulation. Privacy issue will also be looked into with utmost seriousness. The privacy policy of Open AI says that whatever information you share will be used by the company for training purposes. Therefore there is no privacy," said Duggal.

India does not have a dedicated privacy policy. Data protection will be very important for generative AI. The problem is the collection of data. AI platforms will also have to be consonant with Data Protection law.

AI can also help detect corporate frauds. The fundamental question is how do we deal with safety of generative AI platforms. If a chatbot or algorithm is using offensive and abusive language, how do we stop it. 

Lots will depend on intellectual property rights coupled with Artificial Intelligence. The distinction between the IP of humans and machines will have to be settled. The AI and IPR regime will have to be customised and regulated. The chances of Generative AI interfering with IPR will be huge and there will be a tsunami of litigations.

Jurisdiction Of AI Disputes

Duggal then emphasised the necessity of examining the jurisdiction surrounding disputes involving artificial intelligence (AI). According to him, under the Information Technology (IT) Act, entities can be held accountable under Indian law if their services affect users in India. This assertion underscores the significance of determining jurisdiction in cases where AI technologies are involved.

"We have no magical answers to bias. We will have to deal with that. We will also have to deal with consumer protection issue. Population of 140 crore cannot be made guinea pigs. Ethical and moral issues will also be taken into consideration. The technology has to be more resilient, trustworthy and reliable. On 15 August 2023, China became the first country to come up with a law on AI. A lot of work has to be done. The fundamentals have to be focused on. Common minimum regulation has to be done. Intersection of AI with other technologies has to be also appropriately addressed. Generative AI platforms should be prepared for 250 crore fine as per DPDP Act," he concluded.