AI Is Today’s Hottest Technology, So CIOs Need To Learn How To Protect It From Copycats

Companies looking to protect their innovations in artificial intelligence and machine learning face a dilemma. Data and algorithms are essential components to developing products based on these technologies, but it is notoriously difficult to protect them from an intellectual property perspective.

There are two possible approaches here. One is to apply for a patent. Another is to seek trade secret protection. When it comes to A.I. models and algorithms, the latter is the better route. The snag is that most CIOs don’t have experience with this form of intellectual property.

Trade secret protection doesn’t require CIOs to share details about the methods they use, which is a must-do for patents. Instead, it requires them to document that the information is not readily available, has commercial value and that steps are in place to keep it secret.

Federal protections

At the same time, trade secret protection affords companies a pretty robust level of legal protection. If someone steals work processes such as an algorithm, owners can seek damages and halt competing sales by taking action through federal courts in the U.S. Damage awards can run into the tens or hundreds of millions of dollars.

In the largest jury award in 2018 for an intellectual property case, HouseCanary sued Amrock for misappropriating appraisal software utilizing data and machine learning methods HouseCanary had developed. The jury found Amrock guilty of theft of proprietary information and awarded a total of $706 million, including punitive damages of $470 million. The case is under appeal so the final outcome and damages may change.

Trade secrets are used to prevent the theft of information and can overlap with patents and copyrights. Data is often thought of as written work, even if it is in electronic form. Written works are usually covered by copyright, but they don’t work well here since copyrights require some creativity in the production of the work. Data alone doesn’t have that.

Algorithms usually produce a functional aspect, but currently, most of the patent offices of the world consider them as laws of nature due to their mathematical basis and don’t allow them to be patented. CIOs are used to working with functional objects and are naturally drawn to patents. They need to change their perspective and think about data and algorithms instead.

Richard Kemp, a partner at Kemp IT Law, suggests that this change in perspective changes the game. “In the AI space trade secrets are arguably the I.P. [intellectual property] that’s fittest for purpose. This is because data isn’t (as such) patentable or (as facts) copyrightable, but both U.S. and E.U. law explicitly protects data as trade secrets.” In a recent blog post, he lays out details of what companies need to do to protect AI inventions.

Another advantage of trade secrets

One of the most useful things about trade secrets is that they don’t need to be registered to offer protection. Even though this is the case, there are still laws that cover how companies can develop and protect trade secrets. IP expert Eric Goldman has called the U.S. one the biggest IP development in years.

There are requirements for what constitutes a trade secret and how they are asserted. The laws require documentation that provides proof that the information is not readily accessible; that it has commercial value; and that steps are in place to keep it secret. Putting processes in place to establish these can be done inexpensively and often can be worked into existing business best practices.

Patents on the other hand cost between $10,000 to $100,000 to file depending on their length and how many countries are covered. Even filing as few as ten of them can lead to companies spending millions of dollars on patents.

Organizations that do an excellent job of protecting their AI innovations can expect to be successful and well-compensated, since algortihms and the data behind them are some of the hottest new technologies. Learning the steps needed to create trade secrets around them is almost as important as developing the innovations themselves.


Original post:

2 comentários em “AI Is Today’s Hottest Technology, So CIOs Need To Learn How To Protect It From Copycats

Leave a Reply

Your email address will not be published. Required fields are marked *