The UK’s financial regulator is taking a new step forward in the world of artificial intelligence. It has decided to hire Palantir Technologies to analyze sensitive data in a short trial, which may influence the way in which the detection of financial crimes is carried out in the future.
The UK’s Financial Conduct Authority (FCA) has decided to hire Palantir to assist it in detecting fraud, money laundering, and insider trading at the 42,000 firms under its supervision. These firms include banks, lenders, and cryptocurrency exchanges.
Palantir will be using the FCA’s large data lake, which includes case files, reports from firms, and complaints from the public. It will also be using emails, phone call data, and social media data. A significant part of the data is highly sensitive.
The trial will be carried out for three months, and the FCA will be paying more than £30,000 per week to Palantir.
Innovation, Influence, and Data Sovereignty in the UK
Palantir is no stranger to the UK government scene. The company has managed to secure deals worth over £500 million. This includes working with the National Health Service, the police, and the Ministry of Defense. The fact that one company seems to be dominating the government scene in the UK has sparked debate.
The company was founded by Peter Thiel. This person is known for his political affiliations in the United States. This adds to the debate as to why there are concerns regarding the sharing and possible misuse of data insights.
Inside the FCA, there are concerns. There is one major concern: if Palantir were to gain access to the methods the FCA uses to fight financial crimes, could this information be leaked to others?
Experts have mixed opinions regarding the matter. According to Professor Michael Levi from Cardiff University, there are many instances when the data collected by the FCA is not fully utilised. This could be changed with the help of AI. However, there is a major concern: what are the regulations regarding what Palantir can do with the information it collects?
Privacy groups are even more explicit. They say that there are risks involved in inputting real-world data into AI systems. Financial investigations may involve a lot of data. This may even include data from individuals who are not under investigation. Names, bank details, and personal messages may all be included in a file.
The Palantir-FCA Partnership: Security Guarantees vs. Ethical Concerns
The FCA says it has set strict limits. Palantir will be a “data processor” only. This means that it can process data only as per the FCA’s orders. The FCA will be in charge of encryption keys. The data will remain in UK servers. After the contract, Palantir must delete the data.
The FCA also says that Palantir cannot use the data to train its own products. The data analysis or products developed during the trial are to be owned by the FCA.
Nonetheless, the argument that these measures go far enough is debatable by many critics. Some argue that even temporary access can reveal patterns. Others say that best practice recommendations involve the use of synthetic or anonymized data in trials. The FCA chose real data for its trials and argued that it produces more effective results.
Politicians have also had a say in the matter. Members of the House of Commons have previously spoken out against Palantir. They point to the company’s dealings with military and immigration authorities. Some call the company a “controversial company.” They raise human rights concerns.
However, Palantir argues that it has a good record to show for its services. The company claims that it has helped improve the running of hospitals in the NHS and that it has assisted the police in tackling serious crimes. The company also claims that it has a strict policy on human rights.
The FCA’s AI Tightrope in the Fight Against Financial Crime
For the FCA, the aim is clear. The complexity and opaque nature of financial crimes mean that traditional methods do not keep up with the crimes committed. AI provides a solution to scan for patterns and alert authorities to potential crimes.
But what a price to pay. There may be a trade-off between more detection and more access to data. This begs questions about trust, control, and accountability. This trial is at the heart of a broader debate. There is a desire from governments to increase the power available to combat crime.
There is a need to protect citizens’ and companies’ data. This project may inform just how far governments will go in terms of AI. For now, the FCA is simply testing what is possible. What remains to be seen is whether there are real benefits to be gained from AI without crossing a line that society is not ready to cross.
