The volume of data that organisations must deal with in legal disputes and compliance investigations continues to increase exponentially year on year, as the sources of data also evolve rapidly. With the analytical technology and methods for reviewing data quickly becoming outdated and inefficient, businesses with already stretched resources can find it hard to keep up.
Currently, whenever a legal matter or investigation arises, potentially relevant data is collected from different company sources into a single database, where legal teams attempt to weed out irrelevant data before spending thousands of hours reviewing documents one by one, categorising and flagging important or sensitive data. When the next case arises, the whole process is repeated, often with the same data collected, culled and reviewed again, duplicating the cost.
This siloing of information means there is no way of sharing data and learnings across matters. A lawyer in one case may spend weeks reviewing thousands of documents to identify those that contain protected personal information of clients, despite those same documents having previously been reviewed and redacted for that information on a different matter. Such unnecessary work and rework creates inconsistency, plus the potential for sensitive information to be accidentally produced.
The need to protect sensitive information is a real concern to companies with ever-increasing data volumes. The legal and regulatory frameworks that protect personal data are rapidly changing and becoming increasingly stringent, with a growing focus on the General Data Protection Regulation in Europe and, in the United States, recently enacted Californian privacy laws creating similarly strict regulations that other US states are poised to follow.
These regulations bring incredibly high monetary penalties if companies inadvertently disclose personal information and yet, when dealing with unstructured data, it can be difficult to comply.
“It can be hard to prevent sensitive information from being pulled in because organisations do not know it is there,” says Karl Sobylak, senior product manager at Lighthouse, which has been innovating in enterprise data for compliance and legal for 25 years. “For example, an employee emails a co-worker and includes a spreadsheet of birth dates, that spreadsheet then gets into the email system and it is hard to keep it from showing up in databases.”
That spreadsheet containing personal data can get pulled into multiple review databases for different cases, increasing the risk it gets missed and illegally transmitted to an opposing party.
Older generations of artificial intelligence (AI) and analytical technology are ill-equipped to respond. Most courtrooms and legal professionals are familiar with technology-assisted review (TAR) but, as Sobylak explains: “The limitation of most TAR is it can only be used in the context of a single case and requires significant input and turning of data that can take several weeks. Then, when you’re done with one case, you move onto the next and reinvent the wheel. So far, we’ve seen that 42% of documents are reviewed on multiple matters. At the same time, 94% of documents being flagged as potentially privileged are not actually privileged.”
Now, advances in AI and machine-learning can help, increasingly moving on from troubleshooting to adding real value.
Sobylak says more advanced AI can solve the siloed data problem where companies pay for data to be reviewed and re-reviewed. “TAR is all based around a single algorithm,” he says. “This new generation of big data technology analyses text, metadata and prior lawyer decisions using a number of algorithms to give context around who is talking to who, what their communications were like previously and how prior reviewers interpreted that information, creating the ability to do a much better job.”
These advances allow a more holistic approach, aggregating data from hundreds of past matters and using it to spot trends and support data-driven decision-making. Sobylak says: “Not only does this speed up document review and deliver cost-savings, it allows for consistency within and across matters, which can significantly lower a company’s risk.”
This can also be invaluable at lowering the risks associated with sensitive information. Because it can leverage multiple algorithms across previous lawyer work product, this new AI can more accurately find the types of personal information that need to be redacted or withheld. If a document was flagged as containing personal information on one matter, it will be flagged on subsequent ones.
94% of documents being flagged as potentially privileged are not actually privileged
It can also help identify data protected by legal professional privilege, which can require more than the basic identification of keywords and calls for additional context. Newer AI technology is capable of identifying the context in which a conversation between two lawyers takes place and whether it is legally privileged, leveraging knowledge learnt from decisions made on previous matters.
There is the potential to solve many of the big data challenges and drastically improve cost efficiency for companies. Should this worry lawyers, who were previously paid to conduct all those hours of reviews and re-reviews?
Rob Hellewell, vice president of Lighthouse and a former practising lawyer at a large international law firm, says: “AI is not going to replace lawyers. This technology is about enhancing lawyers’ ability to separate the chaff and get to the wheat. Our focus is on augmenting their ability to find relevant, privileged and sensitive information faster, consistently and at less cost.”
Hellewell argues that lawyers often fail to appreciate the data aspects of their roles, but jump on board once they see results. “Imagine there’s a new matter with a million documents to review and you can start by eliminating the need to review ones that have been looked at previously,” he says. “Lawyers can see those benefits immediately.”
In addition to simply making day-to-day tasks easier and work product more consistent, newer AI solutions continue to break new ground and add more intrinsic value to legal practice.
“Normally we are just providing the information that we are required to provide to settle the case. But AI now gives us the opportunity to look at how we might have prevented a dispute from arising or identified a problem sooner,” says Hellewell.
From a company-wide standpoint, the latest AI offers potential to reduce risk and improve cost efficiency, while from a legal team perspective, it can unlock valuable legal insights from company data and enhance the critical contribution of in-house lawyers and external law firms.
For more information please contact info_london@lighthouseglobal.com
Promoted by Lighthouse
The volume of data that organisations must deal with in legal disputes and compliance investigations continues to increase exponentially year on year, as the sources of data also evolve rapidly. With the analytical technology and methods for reviewing data quickly becoming outdated and inefficient, businesses with already stretched resources can find it hard to keep up.
Currently, whenever a legal matter or investigation arises, potentially relevant data is collected from different company sources into a single database, where legal teams attempt to weed out irrelevant data before spending thousands of hours reviewing documents one by one, categorising and flagging important or sensitive data. When the next case arises, the whole process is repeated, often with the same data collected, culled and reviewed again, duplicating the cost.
This siloing of information means there is no way of sharing data and learnings across matters. A lawyer in one case may spend weeks reviewing thousands of documents to identify those that contain protected personal information of clients, despite those same documents having previously been reviewed and redacted for that information on a different matter. Such unnecessary work and rework creates inconsistency, plus the potential for sensitive information to be accidentally produced.
