AI that monitors: when does productivity analysis become an invasion of privacy?

The case recently reverberated on social media in which one of the largest banks in Brazil used technological tools to identify, map, evaluate and dismiss, in masse, employees who worked from home has reignited the discussion about the limits of employers' inspection power, increasingly feasible with Artificial Intelligence (AI) tools versus the right to privacy of workers, since work performed in the home office often mixes personal activities with professional activities.

Brazilian legislation on remote work dates to the 1940s. Article 62, I of the Brazilian Consolidation of Labor Laws (CLT) provides that employees who "perform an external activity incompatible with the establishment of working hours"  do not need to record the working hours.

Doctrine and jurisprudence are already settled in the sense that if it is possible to control working hours, the worker's clocking in cannot be excepted, and the application of item I of article 62 of the CLT must be ruled out.

At the time the law was consolidated, it was not possible to control work remotely, as there were no GPS, cell phones, login systems, and applications.

Not even the 2017 labor reform, when it brought item III to article 62 of the CLT, which deals with remote work (including working from home), imagined the popularization of a technology such as AI.

In this sense, it can be said that with remote work and AI, the foundations of article 62, I of the CLT have been emptied by the new reality. It is almost impossible to imagine any work performed remotely that cannot be monitored by technological means, such as geolocation (GPS) on corporate cell phones, remote login/logout software, AI systems that track connection time, use of productivity applications, among others.

In other words, these technologies make control of working hours and productivity fully feasible, even if the employee is not physically in the company, which requires the conclusion that the new reality ended up emptying article 62, I of the CLT - not by law, but by incompatibility with the technological reality.

Several AI tools are currently used by companies to control employees' activities, schedules, productivity, attendance, compliance with rules, and achievement of goals.

These tools help to give visibility and generate insights into productivity and telecommuting paradigms, being able to detect behavior patterns that are out of the average or identify where there are bottlenecks.

So much technology available to companies makes us ask: what about the issue of workers' privacy? Would it be the end of autonomy in labor relations? To what extent can the search for transparency and ethics in labor relations invade the privacy of workers, exposing data of the subjects of the labor relationships who have constitutional rights such as human dignity, the right to privacy and intimacy?

The Labor Court has always been very protective and strict with companies when it comes to the right to privacy and intimacy.

The protective principle of labor law has always been applied in the Labor Courts, especially when there was doubt. It is often said that it is the principle "in dubio pro operário", that is, if there is doubt about the interpretation of a rule, the interpretation must be favorable to the worker.

It turns out that the technological transformation that the labor relationship has been undergoing is already being felt by the judiciary. In this scenario, AI is being seen as an important tool that ends up reducing uncertainty and bringing evidence of what was once doubt.

Topics such as equal pay, in which the company used to have difficulty proving that "employee A" produced less than "employee B", can now be easily proven by companies that have productivity control tools.

The Labor Court has addressed the issue of monitoring remote work in several courts. Among the most recurrent examples, the use of its own monitoring applications with control of working hours by communication platforms and login/logoff records on the company's equipment, among others, stand out.

In all these cases, the focus of monitoring falls on productivity and the execution of tasks by employees, usually with the objective of verifying the compatibility between the recorded working hours and the effective performance of activities.

However, Bill No. 2338/2023 (PL) - which provides for the development, promotion and ethical and responsible use of AI in Brazil, currently in progress in the Chamber of Deputies - classifies as high-risk systems those tools that use Artificial Intelligence for purposes related to the management of labor relations and evaluation of employee’s performance and behavior. This classification stems from the potential of these systems to impact the fundamental rights of workers.

The text does not prohibit the use of AI in these contexts, but imposes reinforced governance obligations, such as the need for transparency, meaningful human oversight, explainability of automated decisions, and mitigation of biases and discrimination.

In practice, this means that companies that adopt AI tools for people management must ensure that such systems are auditable, documented, and used in a proportionate and ethical manner, with respect to the Brazilian General Data Protection Law (LGPD) and current labor legislation.

In short, the incorporation of Artificial Intelligence in remote work management has brought evident gains in efficiency and control, but it has also revealed a sensitive limit between monitoring productivity and invading privacy. Care is necessary to prevent the promise of optimization from giving way to constant vigilance - with risks to autonomy, well-being and trust in work relationships.