**6. Artificial intelligence and privacy**

Artificial Intelligence (AI) is a broad field that includes machine learning and cognitive computing where computers are programed to mimic human cognitive functions such as learning and problem solving but many times, much faster and in more accurate ways [14]. The use of AI is expanding into a plethora of areas including speech recognition, facial recognition, medical diagnosis, financial predictions, tracking of disease outbreaks etc. AI algorithms enable computing systems to rationalize and take actions aimed at achieving a specific goal or set of goals.

User and stakeholder security can be enhanced through AI tools, which can take advantage of blockchain to open up new avenues for accessing and learning from data without taking ownership or control of that data. This can reduce risk for the organization and the stakeholders who provide the data. Both individual blockchain members and the organization or group in charge of setting governance rules and processes can benefit from building in privacy-related AI functionality (as early as possible) in the design of blockchain networks and processes.

Companies have implemented AI to create holistic views of customers by piecing together transactions from all customer touchpoints. Blockchain participants will have incentives to pull together integrated datasets by combining transactions for a single customer across all blockchain partners. This creates potential benefits for blockchain partners, but can also negatively affect the privacy of customers and other stakeholders for which this integration is possible.

In combination with options for identity protection through decentralization, AI can be used to combine personal data from blockchain participants and their stakeholders in a way that maintains information security and personal data privacy. Through these processes, user and stakeholder security can be enhanced and data sets and AI models can be improved.

We can identify four categories of stakeholders that can be affected by an organization's data transparency and privacy processes: (1) participants, whose data—both direct and indirect—are gathered; (2) victims, who are affected by decisions made using participant data; (3) users, who use participant data in their work; and (4) custodians, who manage and secure data. When AI can be used to manage access to data and to develop analytical models using that data, all stakeholders can benefit [15].

**Table 1** summarizes a number of ways AI can be used in a blockchain setting to protect or increase privacy of user's personal data. This AI/Blockchain combination can increase system security by helping to detect attacks by bad actors, user security by sharing permissions and smart contracts, enable privacy-enhanced use of datasets through improved identity management and better data, and it can improve AI models through more varied, valid, and ethically-sourced data and better hypotheses. Each item in **Table 1** is described briefly below, followed by examples of use cases using this combination of technologies.

Computational intelligence (CI), a subset of AI can improve the Blockchain's attack resilience thus improving security of the system and ultimately the privacy of the data residing on the system. AI is rooted in hard computing techniques whereas computational intelligence is based on soft computing methods, which enable adaptation to a range of changing variables [16].

Computational intelligence, when combined with blockchain systems, can create more robust cryptographic functionality and ciphers thereby making it more difficult for cyber hackers to compromise systems even as computing power and efforts to hack these systems over time increases. Quite appropriately, [14] refer to the intersection of blockchain and AI as "blockchain intelligence". Additionally, AI algorithms can be built on blockchains to detect when a blockchain is under


**Table 1.** *The role of AI in blockchain user privacy.*

#### *How Blockchain and AI Enable Personal Data Privacy and Support Cybersecurity DOI: http://dx.doi.org/10.5772/intechopen.96999*

attack by continually monitoring blocks and activity on the chain. This technology increases trust in the system beyond what the native architecture provides [17, 18].

When blockchain participants have increased control over their own data, they have the potential to decide with which parties and for what purposes their data are shared. In order to collect participant data for use in an AI dataset, participant permissions will need to be obtained. This provides users with 'opt-in' control, rather than 'opt-out' and helps to ensure that personal data is used in ways that are consistent with the intentions of the owner. In some cases, which may become increasingly common if decentralized identity solutions are adopted, users can be compensated for providing their data to organizations seeking to utilize this data in traditional and AI decision models.

Smart contracts can also protect privacy. Permissions granted by users can be subject to complex rules embodied in smart contracts, which can enforce rules regarding the use of the data and can govern the granting and rescinding of participant data. AI can be used to scan contracts to identify participants who have or are likely to possess or provide data for desired uses.

The size and nature of datasets available in blockchain networks can also have implications for effective AI. Because many different organizations and stakeholders contribute to shared ledgers, the quantity and variability of data available for analysis can be much larger than for single-company databases. Larger datasets could enable more sophisticated identity-masking procedures, and metadata may be richer and more informative.

Because of the validation, security, timestamping and the append-only nature of blockchain ledgers, the data obtained are likely to be much cleaner and more accurate than when data are captured and maintained by many organizations in databases that are not immutable.

The ethical quality of data obtained will also be higher, and model developers and users can have increased confidence that they are following regulations. Because multidimensional user permissions can be granted and documented—and in some cases, enforced through smart contracts—organizations can use this data with less risk of privacy breaches. In addition, because user data can be collected using zero-knowledge proofs, complex analyses requiring specific user data can be performed and the necessary information captured and used without the need for accessing or possessing PII.

The use of blockchain data and artifacts can also result in higher quality analyses and outcomes. When data are clean and associated with clear metadata, the validity of the data is increased. Because each item in a data set is more trusted, error can be reduced and insights can be obtained through smaller data sets. When clean data are used to train AI models, those models will be more accurate, and the predictions and decisions made by those models will also be improved. Clean-training data can also be useful in validating non-blockchain data for use in AI models.

Finally, and perhaps most importantly, the hypotheses upon which AI models are built can improve, for several reasons. First, because participant permission must be obtained and possibly paid for, AI designers will need to develop clear designs that define the analyses to be performed and determine the type and amount of data needed for these analyses. This will require designers to be more aware of the universe of data that could inform these analyses and what is and is not available in distributed ledgers and personal-data files. This could help identify problems such as the lack of black faces in photo-categorizing algorithms before or during data collection.

PII may never be collected, and when it is, its use may be more intentional and usage agreements may be enforced by smart contracts. This enables more ethical approaches to gathering and managing data. AI models built using ethicallysourced and governed data can generate results that are actionable within predefined ethical and regulatory limits.

#### **6.1 Emerging blockchain and AI industry uses cases**

The 2019–2020 Covid 19 pandemic has prompted medical researchers and technologists to research ways to quickly gather intelligence around virus exposure and transmission as a way to combat the spread of the disease while maintaining personal privacy of users. Point-of-care diagnostics, which rely on rapid testing of patients that may have been exposed to the virus is proving to be an effective way of tracking the spread and reducing the impact of the disease. German based Pharmact AG has developed a rapid Covid 19 test that delivers results in roughly 20 minutes. This test can be used in point-of-care systems and combined with blockchain and AI to increase the speed of diagnosis and provide statistics on positive and negative results while maintaining security of personally-identifiable data. Data can be collected on blockchain infrastructure while taking advantage of the speed that AI affords to create an integrated platform that enables data from disparate sources to be analyzed. Information drawn from these systems can provide communities a powerful tool for combatting the spread of disease, reducing the burden of health care facilities and saving lives [19].

Many cities are working toward becoming "smart cities" by integrating AI and blockchain with other web 3.0 technologies such as internet of things (IoT) sensors and edge devices. Intelligent transportation systems are enabled by these technologies. Self-driving cars make use of IoT sensors to continuously monitor surrounding situations and even anticipate developments by using artificial intelligence. These cars can incorporate blockchain wallets that enable passengers to pay for rides, rentals, tolls, etc. without revealing personal data. By adding blockchain as an underlying architecture, cities and private companies can reduce the friction of renting or sharing autonomous vehicles by streamlining the process of procuring a ride. The peer to peer nature of blockchain reduces the number of people or businesses involved in the process, taking out expensive intermediaries, and reducing costs. These systems can also provide audit trails for both owners and renters, and enable rating and payment systems that maintain privacy for both parties. Data gathered by the vehicles can contribute to learning algorithms on the blockchain for increased security, scalability and efficiencies as well as improved transportation and sustainability for the city [20].

Smart home systems that preserve user privacy while contributing usage data for analysis can likewise benefit from the integration of blockchain and AI. Smart home systems are becoming popular and manufacturers increasingly enable connectivity between devices. These systems are valuable sources of consumer usage data. AI-enabled blockchain systems can be used to push machine learning and training processes to consumer's mobile devices and edge computing servers. Users can then submit locally-trained models for analysis, in some cases with option of adding noise that makes it very difficult to trace shared data back to individual consumers. Decentralized technologies enable analysis of locally generated data without this data being submitted to a centralized server [21].

These use cases exemplify some of the ways blockchain and AI are being used to accomplish objectives while maintaining personal data privacy. New use cases continue to be developed as technologists and user communities recognize the possibilities for systems that provide both functionality and privacy.

#### **7. Conclusion**

Blockchain and AI technologies are improving at a rapid pace and enabling possibilities for sharing and combining data in ways not previously envisioned.

#### *How Blockchain and AI Enable Personal Data Privacy and Support Cybersecurity DOI: http://dx.doi.org/10.5772/intechopen.96999*

At the same time, advances in these technologies provide new possibilities for the ethical use of data. Personal data, when shared, present a conundrum for firms and individuals, which can provide valuable benefits but can also create great risks and costs for both the individual and the organizations with which individual data are shared. Blockchain provides new mechanisms, such as decentralized identities and zero-knowledge proofs, that enable data to be shared in ways that maintain the privacy of the individual and allow users to maintain control over their own data. These advances can provide both increased cybersecurity and more ethical use of personal data. Blockchain participants can realize these outcomes through careful development of governance frameworks and mechanisms.

Publication of this chapter in an open access book was funded by the Portland State University Library's Open Access Fund.
