Luckily, artificial intelligence in its literal sense is just a buzzword. The terminology “Artificial Intelligence” was born during a Conference in Dartmouth 1956. Since then computer technology has made giant steps. But, until today the technical means to create intellectual capability matching those of humans does not exist. Hence, one differentiates between “weak AI” – until today mostly limited to self-optimizing pattern matching software recognizing pictures, written and oral text mimicking aspects of human problem solving – and non-existing “strong AI” – capable to match with the human intellect -. The existing “weak” AI is already applied in many business sectors, e.g. the digital health, the automotive as well as the financial sector.

The WIPO addressed this hot topic under the title: “WIPO Conversation on Intellectual Property and Artificial Intelligence” in an intergovernmental as well as stakeholder meeting on September 27, 2019 (see documents and references https://www.wipo.int/meetings/en/details.jsp?meeting_id=51767) The WIPO has convened the meeting to provide member states with an opportunity to exchange views on various topics regarding AI and the IP systems. The meeting focused on the impact of AI on IP systems, IP policies, IP rights management, and international cooperation on IP matters. AI, big data analytics and new technologies such as blockchain were considered as the growing challenges facing IP offices (IPO).

According to all economical publications around the globe, AI based business models are already or will be the game changer in nearly all relevant markets with an estimated economical effect of 15 Billion EUR until 2030. Hence, venture capital investors as well as public funding agencies are strongly attracted by AI applications, while asking for ways of Intellectual Property (IP) protection with respect to those AI applications.

Basically, when addressing the issue of AI protection, we legally deal with IP based on Copyright as AI technology is basically an algorithm written in a code. Besides, a system of Trade Secret protection is widely used. Last, but not least, AI technology might be patented as an invention with an embedded software, as in the EU software as such is excluded from patent protection.

Copyright

Software developers generally rely upon copyright as the sole form of IP in their code. This is mostly driven by the fact that copyright comes into existence automatically, without any of the costly and time-consuming formalities of any other IP protection.

This causes a misconception to the effect that small and medium sized companies as well as start-up companies are hesitant to invest in IP protection, until third parties are asking for proof of IP in exchange for funding the venture.

Unfortunately, in the EU software as copyright can not be protected by filing in a public register. Nonetheless, a company doing business in other Non-EU markets should take legal advice if in those other jurisdictions software can be protected by registration, e.g. as in China.

Besides, it´s worthwhile to have a look at the EU Database Directive (Directive 96/9/EC of the European Parliament and of the Council of 11 March 1996 on the legal protection of databases). The EU Database Directive protects non-original databases if the investment in obtaining, verifying and presenting the data was substantial. As AI technology is generally based on substantial investments such as “deep learning” methods the database protection will apply also to AI technology as it is often using large databases obtained from third parties or based on single data collected from multiple third parties. That´s why this right of database protection is considered a “sui generis” right. The EU Database Directive applies especially to digital databases.

The discussion if AI technologies create a new authorship and in how far AI must be considered as a legal personality seems to be on hold, at least in the EU (see EU Expert Group on Liability and New Technology https://ec.europa.eu/digital-single-market/en/artificial-intelligence )

In consequence companies are well advised to protect AI technologies under the EU Database Directive and make it less vulnerable to infringements by using the appropriate legal measures of database protection.

Trade Secrets

A new cross-border tool to stop infringements in the EU was introduced in the process of harmonization by the Trade Secret Directive (Directive (EU) 2016/943 of the European Parliament and of the Council of 8 June 2016 on the protection of undisclosed know-how and business information against their unlawful acquisition, use and disclosure).

The directive introduces an EU-wide definition of “trade secret”, meaning information which is secret, has commercial value because it is secret, and has been subject to reasonable steps to keep it secret. Furthermore, it obliges EU member states to ensure that victims of infringements of trade secrets are able to defend their rights in court and to seek compensation.

In principal AI technology is a classical example for a “trade secret” generally not disclosed to the pubic.

The crucial element of legal protection is the investment in reasonable steps to keep the AI technology code/algorithm and database secret. This requires companies to have robust procedures, tight contracts and strong cybersecurity measures in place, ideally set up and monitored by the developers in cooperation with legal advisers

In consequence companies are well advised to protect AI technologies under the EU Trade Secret Directive by setting up a strong protection scheme legally advised in order to be able to enforce the rights in case of infringement.

Patents

Finally, the registration of an invention patent seems to be the most wanted solution to protect AI technology.

As in the EU software as such is excluded from patent protection, there are still ways to seek protection by filing an invention that is a software based technical solution. In practice, the biggest obstacle to patenting a software based technical solution usually lies in the large body of prior art against which such patent applications are judged by patent offices.

Besides, one should keep in mind if the business case extends beyond the EU to other jurisdictions, patent protection for AI technologies might be granted more easily- e.g. as in the US and China supported by the high number of AI patent filings and registrations with the USPTO and the CNIPO -.

Furthermore, one should note that it is contemplated if AI based inventions create a new legal AI inventorship – a topic formally addressed by IBM during the WIPO conference on September 27, 2019 with reference to a Federal Register Notice by the USPTO seeking comments on AI inventions/patents, with a deadline October 11, 2019.

In consequence companies are well advised to consider patent protection with respect to all potential markets in which the AI technology might be used.

Conclusion

AI technology needs IP protection as it protects the time and money invested in the process of innovation, keeping the risk of infringement low and the likelihood of third party funding high.

But, there is no one-size-fits-all IP approach. Copyright, Trade Secrets and Patent protection regimes are functioning as multiple layer of protection and are likely to serve the purpose of effectiveness. But, as enforcement is crucial and AI technologies are likely to have a cross-border market, the legal advice must cover all relevant markets right from the start.

Prof. Dr. Frank HAMMEL Partner WEITNAUER

Prof. Dr. Frank HAMMEL is partner and certified Lawyer for Intellectual Property at WEITNAUER in Berlin. He advises nationally and internationally operating companies in the field of IP / IT – with a focus on digital change, Industry 4.0, IoT – as well as in managing international brand portfolios.