The Government is rolling out artificial intelligence (AI) technologies as part of an effort to clamp down on the risk of fraud in the immigration process.
A briefing document drawn up for Minister for Justice Jim O’Callaghan in February outlines that the Department of Justice has begun using a system called ID-Pal as part of its self-service digital contact centre, a portal set up for people using immigration services.
The ID-Pal system is used for identity verification purposes when users are setting up an account. It uses AI and machine learning to “verify in real-time the authenticity of applicants’ identity documents, ensuring that the person is accurately identified, reducing the risk of fraud”.
The tool uses “liveness detection” alongside “facial matching and document verification” to confirm the person providing the information “is present, is the true owner of the documentation and that the documentation is authentic”.
READ MORE
It does not form part of the assessment of any immigration or international protection application, a spokesman for the department said.
The department also uses two public-facing chatbots within the immigration service and international protection website to answer queries drawing on material uploaded by the department.
The AI chatbot is used to provide answers to the public, in the hope that it will reduce the number of queries and free up staff to work on “complex issues that require human interaction”, according to the document.
A third AI tool is used to provide customer support, although queries can be submitted for response by staff. This tool also has a function for staff, which officials can use to find Department of Justice materials for more complex queries as well as automating simple tasks like responses or gathering multiple queries from one person.
Final responses are issued by civil servants. It is not used for decision-making, according to the document.
The department has said the rollout of AI is governed by safeguards and training for staff. It has added that the rollout complies with Government guidelines, guidance from the National Cyber Security Centre and with the EU Artificial Intelligence Act.
Experts cautioned that its use must be closely monitored, especially when it is deployed in an immigration setting.
TJ McIntyre, a privacy specialist and assistant professor in the UCD School of Law, said: “You are dealing here with people in a vulnerable situation.” He said that these people, in many cases, do not speak English as a first language. “In light of all that, you have to treat it with huge care,” he said.
With regard to the use of AI to verify documents, he said this would rely on the processing of biometric data such as photographs of a person, which needs to be done on a particular legal basis and with specific safeguards in place under the EU’s General Data Protection Regulation. Dr McIntyre also warned that the system must be open to review by non-governmental organisations and immigration lawyers to see how they operate.
Solicitor Ángel Bello-Cortes, a partner with Fragomen, which specialises in immigration matters, acknowledged that AI is a “powerful tool when used properly” but cautioned against anything that might “insert AI into the decision-making process”.
Any reform of the immigration process should focus on developing “clear, comprehensive and transparent” guidance for applicants, “consistent and timely decision making, particularly for entry visas and family reunification”, and “meaningful engagement” between justice officials and applicants through their online portals.
“AI could be helpful in supporting, internally, these structural improvements, but decision-making itself must remain entirely in the hands of human officials,” said Mr Bello-Cortes.
The department said its use of the chatbot is low-risk and it is not considering high-risk AI systems at this time.
