Asset 1
When Speed Brings Security: The Panama Papers Investigation

The Panama Papers disclosure is justifiably acclaimed as triumph for the pioneering field of data journalism. Yet, behind the stunning exposé of how rich and powerful people hide their money, is the story of a data investigation so complex and incendiary that it endangered the lives of those working on it. Now, artificial intelligence is helping to accelerate financial crime investigations where speed and secrecy are paramount.

“I like to refer to Johannes Kristjansson, the Icelandic reporter we invited to join the project, as the loneliest man in the world,” said Gerard Ryle, the leader of the international team that investigated the Panama Papers. Speaking to TED, he recounts how the journalist devoted himself to the project for 9 months, refusing all other work and living off his wife’s income. Working late into the nights, he kept the windows of home covered to ensure no one could spy on information that would ultimately bring down the leader of his country.

Kristjansson’s self-enforced seclusion hints at the threat faced by the 3 76 journalists who collaborated to probe the world’s largest ever data leak. Their remarkable endeavor was set in motion when an employee of Mossak Fonseca, a Panama-based law firm, passed the first tranche of 11.5 million documents to two journalists working for Süddeutsche Zeitung, a German newspaper. Reaching back almost 40 years, the cache of information linked politicians, heads of state, business leaders and their associates to tax evasion, smuggling, terrorism, money laundering and organized crime.

Faced with an investigative task well beyond their resources, Süddeutsche Zeitung sought the support of The International Consortium of Investigative Journalists (ICIJ), based in Washington, DC. In what Ryle describes as “the biggest journalism collaboration in history,” 100 media organizations from 76 countries cooperated in absolute secrecy.

The biggest kind of silence

The Panama Papers contain documents and communications related to 214,000 shell companies, many of them created to hide and move illegal funds. As the investigation progressed, the journalists uncovered dozens of revelations. Many could have made headline news on their own. However, as he corralled his team of reporters to fight the instinct to publish and focus on the bigger picture, Ryle knew that “in order to make the biggest kind of noise, we first needed the biggest kind of silence.”

That silence masked a complex operation supported by sophisticated technology. The file size of the Panama Papers dwarfed those by WikiLeaks or Edward Snowden. A total of 2.6 terabytes of data (1 terabyte equates to 1,000 copies of the Encyclopedia Britannica) was spread across 5 million emails, 2 million PDFs, and millions more files and documents. All that information, most of it in unstructured text written in dozens of languages, needed to be made readable and searchable.

Working with multiple vendors, the ICIJ rendered the huge volume of unstructured data into an indexed database. A secure virtual newsroom was created, using a specially designed search engine and an encrypted communication systems that allowed the team to share and discuss the information emerging from the documents. A phalanx of domain experts was recruited to the task, each contributing their knowledge of national regimes, criminal networks, and arcane matters of tax, law and finance. As insights began to emerge, graph database and data visualization software was used to map links between entities so that, bit by bit, the complex webs of offshore tax regimes, financial intermediaries, shell corporations, and beneficial owners could be gradually revealed.

Vitriol, cyberattacks and lawsuits

Even with 376 journalists at work, the investigation took more than a year. Throughout this time there was the ever-present danger of hacks from criminals or state-sponsored organizations like Russia’s FSB, or internal leaks either due to carelessness or coercion. Maintaining secrecy helped to protect the journalists’ lives and livelihoods. For months before and after the project’s release, reporters were assigned armed bodyguards posing as Uber drivers. Ryle also admits the group faced numerous crises, as editors itched to publish or became nervy about the investigation being exposed.

As we know, the Panama Papers made global news and dealt a massive blow to money launderers and their clients. The backlash against some of the journalists showed why tight security was vital. Implicated politicians, business leaders and their supporters responded with vitriol, cyberattacks and lawsuits. Personal information about the journalists, such as addresses and details about their children, was shared online. Several were sacked from their jobs. Some even faced jail and death threats. But now the world knew. By exposing how the rich and powerful use corporations and the global financial system to shield their wealth, a courageous band of journalists has paved the way for new legislation that will make financial crime harder to commit and easier to prosecute.

The baton has now been passed to financial institutions and regulators. The Panama Papers will go down in history as astonishing journalistic endeavor, but it has also highlighted the perils of handling injurious information about corrupt elites. Compliance organizations and financial crime investigators are at war with powerful forces. Money launderers methods will evolve and their motivation to bribe or coerce insiders will increase. The combined impact of new laws, tougher scrutiny and the potential for asymmetric threats puts institutions under pressure to accelerate complex investigations.

Precision and speed

Around the time that the ICIJ was bringing together its team, law enforcement in the United States was starting to use artificial intelligence (AI) to explore massive online data sources that could help them identify human traffickers and their victims. The technology is built into a dedicated investigation tool called Spotlight, developed by Digital Reasoning and Thorn. It helps officers scrutinize the 100,000 escort ads posted daily on the Internet in the United States. Within this data are children who are bought and sold online for sex.

Spotlight offers investigators two key advantages: precision and speed. Trained with the domain knowledge of experienced human trafficking investigators, the technology zeros in on the ads which are for children by learning indicative language, connecting disparate data sources, and providing deep investigative tools that help to clarify the historical and full geographical reach of a victim’s trafficking situation. This ability to make sense of human communications with human-like acuity but at vast scale has enabled those law enforcement organizations which use Spotlight daily to achieve a 60% time saving in their process. As one investigator in New York put it: “I get results fast, allowing me to gather important evidence and information in minutes compared to hours.”

Another factor in complex investigations is that the data will often change while the task is ongoing. As reporter at one the newspapers in the Panama Papers investigation commented: “As we worked, more information was being leaked.” New details may yield greater discoveries or more robust evidence, but the difficulty is how to usefully weave these into an already complex sets of facts. For officers using Spotlight this isn’t a problem. Constantly vigilant and with machine learning capabilities that automatically match relevant insights to profiles of entities, the tool accumulates knowledge about victims and perpetrators making it increasingly effective at disclosing identities, hidden relationships and geolocations.

Bad news for money launderers

If the type of AI used in Spotlight had been available to the Panama Papers investigation it would have been a game changer. Steaming through the terabytes of data, indexing the contents, profiling the entities, and visualizing their connections, it could have achieved in hours what took months for investigators using keywords, lexicons and network graphs.

These gains in accuracy and velocity are coming to be regarded as essential as institutions and regulators double down on financial crime. The bad news for money launderers is that the AI which powers Spotlight is already in use in many of the world’s largest banks. So far it has been focused on regulatory compliance, but the lens is widening to encompass know-your-customer, anti-money laundering and financial crime investigation processes. It would be foolish to think that the crimes of corrupt elites can be eradicated, but thanks to the fearless efforts of the Panama Papers journalists, beefed up laws and scrutiny, and the advent of AI analytics technology, the fight back has never been stronger.