With the rapid evolution of Artificial Intelligence, the regulation of AI usage has become a central and urgent issue in Brazil. The new technology brings immense potential to innovate and transform various sectors, but also raises critical questions about ethics, transparency, and governance. In the Brazilian context, where digital transformation is advancing at a rapid pace, finding the balance between innovation and appropriate regulation is essential to ensure sustainable and responsible AI development.
In an exclusive interview, Samir Karam, COO of Performa_IT, offers an in-depth analysis on the challenges and emerging solutions in AI regulation, highlighting the importance of balancing innovation and ethics in the technology sector.
The regulation of AI in Brazil is still in the structuring phase, which brings both challenges and opportunities.“On the one hand, regulation creates clearer guidelines for responsible technology use, ensuring principles such as transparency and ethics. On the other hand, there is the risk of excessive bureaucracy, which can slow down innovation. The balance between regulation and the freedom to innovate is essential for Brazil to remain competitive in the global scenario,begins Samir Karam, COO of Performa_IT – companyfull service providerof technological solutions, a reference in digital transformation and artificial intelligence.
Shadow AIandDeepfakes: Risks and Solutions
One of the most troubling concepts discussed by Samir Karam is that of “shadow AI”, which refers to the use of artificial intelligence within an organization without proper control or supervision. This practice can lead to various problems, such as data leaks, biased decisions, and security risks.
For example, imagine a marketing team using an AI tool to analyze consumer behavior without approval from IT orcompliance. In addition to exposing the company to legal risks, the unregulated use of this technology can result in the inappropriate collection and analysis of sensitive data, violating users' privacy.
Another scenario is the development of AI algorithms for hiring decisions, which without adequate supervision can reproduce unconscious biases present in the training data, resulting in unfair and discriminatory decisions.
Just like in the case of deepfakes, where created videos or audios use artificial intelligence to manipulate a person's images, sounds, and movements, making it appear as if they are saying or doing something that, in reality, never happened. This technology can be used maliciously to spread misinformation, commit identity fraud, and damage individuals' reputations.
The solutions for theshadow AIanddeepfakes are moving towards creating robust AI governance policies, according to Samir Karam, COO of Performa_IT:
These policies include the implementation of frequent audits to ensure that AI practices are aligned with the organization's ethics and transparency guidelines. Furthermore,The use of tools that detect unauthorized activities and continuously monitor AI systems is essential to prevent abuses and ensure data security.
Samir emphasizes that without these measures, the uncontrolled use of AI can not only undermine consumer trust, but also expose organizations to severe legal and reputational repercussions.
Fake Newsand ethical challenges in AI
The dissemination offake newsAI-generated content is another growing concern."THECombating AI-generated fake news requires a combination of technology and education.Automated verification tools, identification of synthetic patterns in images and texts, as well as labeling of AI-generated content, are important steps. But alsowe need to invest in raising public awareness, teaching them to identify reliable sources and question dubious content”, says Samir.
Ensuring transparency and ethics in AI development is one of the pillars advocated by Samir. He emphasizes thatSome of the best practices include adopting explainable models (XAI – Explainable AI), independent audits, using diverse data to avoid biases, and creating AI ethics committees.
One of the main cybersecurity concerns associated with AI includes sophisticated attacks such asphishing- an attack technique in which criminals attempt to deceive individuals into revealing confidential information, such as passwords and banking data, by impersonating trusted entities in digital communications. These attacks can become even more sophisticated when combined with AI, creating personalized emails and messages that are difficult to distinguish from real ones. To mitigate these risks, Samir suggests thatéFundamentally invest in AI-based detection solutions, implement multi-factor authentication, and ensure that AI models are trained to detect and mitigate manipulation attempts.
Collaboration for Effective AI Policies
Collaboration between companies, governments, and academia is vital for the formulation of effective AI policies. Samir emphasizes thatAI impacts various sectors, so regulation needs to be built collaboratively.Companies bring the practical vision of technology use, governments establish security and privacy guidelines, while academia contributes with research and methodologies for safer and more ethical development.
The multifaceted nature of artificial intelligence means that its impacts and applications vary widely across different sectors, from healthcare to education, including finance and public safety. For this reason, the creation of effective policies requires an integrated approach that considers all these variables.
CompaniesThey are fundamental in this process, as it is they who implement and utilize AI on a large scale. They provideinsightsabout market needs, practical challenges, and the latest technological innovations. The contribution of the private sector helps ensure that AI policies are applicable and relevant in the real-world context.
GovernmentsIn turn, they are responsible for establishing guidelines that protect citizens and ensure ethics in the use of AI. They create regulations that address issues of safety, privacy, and human rights. Furthermore, governments can facilitate collaboration among different stakeholders and promote funding programs for AI research.
AcademiaIt is the third essential piece in this puzzle. Universities and Research Institutes provide a solid theoretical foundation and develop new methodologies to ensure that AI is developed safely and ethically. Academic research also plays a crucial role in identifying and mitigating biases in AI algorithms, ensuring that technologies are fair and equitable.
This tripartite collaboration allows AI policies to be robust and adaptable, addressing both the benefits and risks associated with the use of the technology. A practical example of this collaboration can be seen in public-private partnership programs, where technology companies work together with academic institutions and government agencies to develop AI solutions that adhere to safety and privacy standards.
Samir emphasizes that without this collaborative approach, there is a risk of creating regulations that are disconnected from practical reality or that inhibit innovation.“It is essential to find a balance between regulation and freedom to innovate so that we can maximize the benefits of AI while minimizing the risks,”concludes.
Myths of Artificial Intelligence
In the current scenario, where artificial intelligence (AI) is increasingly present in our daily lives, many myths and misunderstandings arise about its functioning and impact.
To clarify, demystify these points, and conclude the interview, Samir Karam answered several questions in a ping-pong format, addressing the most common myths and providinginsights valuable insights into the reality of AI.
- What are the most common myths about artificial intelligence that you encounter and how do you dispel them?
One of the biggest myths is that AI is infallible and completely impartial. In reality, it reflects the data it was trained on, and if there are biases in that data, the AI may reproduce them. Another common myth is that AI means complete automation, when in fact, many applications are just decision-making assistants.
- Can AI really replace all human jobs? What is the reality about this?
AI will not replace all jobs, but it will transform many of them. New functions will emerge, requiring professionals to develop new skills. The most likely scenario is a collaboration between humans and AI, where technology automates repetitive tasks and humans focus on what requires creativity and critical judgment.
- Is it true that AI can become conscious and take over humanity, as we see in science fiction movies?
Today, there is no scientific evidence that AI can become conscious. Current models are advanced statistical tools that process data to generate responses, but without any form of cognition or own intention.
- Are all artificial intelligences dangerous or can they be used for harmful purposes? What should we know about this?
Like any technology, AI can be used for good or for evil. The danger is not in AI itself, but in how it is used. Therefore, regulation and responsible use are so important.
- There is a perception that AI is infallible. What are the real limitations of artificial intelligence?
AI can make mistakes, especially when trained with limited or biased data. Furthermore, AI models can be easily fooled by adversarial attacks, where small manipulations in the data can lead to unexpected results.
- Is AI just a passing fad or is it a technology that is here to stay?
AI is here to stay. Your impact is comparable to that of electricity and the internet. However, its development is constantly evolving, and we will still see many changes in the coming years.
- Are AI systems truly capable of making completely unbiased decisions? How can biases affect algorithms?
No AI is completely impartial. If the data used to train it contains bias, the results will also be biased. The ideal is for companies to adopt bias mitigation practices and conduct regular audits.
- Do all AI applications involve surveillance and collection of personal data? What people should know about privacy and AI?
Not all AI involves surveillance, but data collection is a reality in many applications. The most important thing is that users know what data is being collected and have control over it. Transparency and compliance with laws such as the LGPD (General Data Protection Law) and the GDPR (General Data Protection Regulation of the European Union) are essential.
