With the accelerated evolution of Artificial Intelligence, the regulation of the use of AI has become a central and urgent topic in Brazil. The new technology brings immense potential to innovate and transform various sectors, but also raise critical questions about ethics, transparency and governance.In the Brazilian context, where digital transformation advances at a rapid pace, finding the balance between innovation and appropriate regulation is fundamental to ensure a sustainable and responsible development of AI.
In an exclusive interview, Samir Karam, COO of Performa_IT, offers an in-depth analysis of the challenges and solutions emerging in AI regulation, highlighting the importance of balancing innovation and ethics in the technology sector.
AI regulation in Brazil is still in the structuring phase, which brings both challenges and opportunities. “On the one hand, regulation creates clearer guidelines for a responsible use of technology, ensuring principles such as transparency and ethics. On the other, there is the risk of excessive bureaucratization, which can slow down innovation. The balance between regulation and freedom to innovate is essential for Brazil to remain competitive in the global scenario”, starts Samir Karam, COO of Performa_IT & AG full service provider technological solutions, reference in digital transformation and artificial intelligence.
Shadow AI's and Deepfakes: 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 leakage, biased decisions and security risks.
For example, imagine a marketing team using an AI tool to analyze consumer behavior without approval from IT or compliancein addition to exposing the company to legal risks, unregulated use of this technology may result in improper collection and analysis of sensitive data, violating the privacy of users.
Another scenario is the development of AI algorithms for hiring decisions, which without proper supervision can reproduce unconscious biases present in the training data, resulting in unfair and discriminatory decisions.
As in the case of deepfakes, where videos or audio created use artificial intelligence to manipulate images, sounds and movements of a person, making it seem to say or do something that, in reality, never happened. This technology can be used maliciously to spread misinformation, defraud identities and cause damage to the reputation of individuals.
Solutions for the shadow AI's and deepfakes they are moving towards the creation of robust AI governance policies, according to Samir Karam, COO of Performa_IT:
“These policies include implementing frequent audits in order to ensure that AI practices are aligned with the organization's ethics and transparency guidelines it is essential to use tools that detect unauthorized activity and continuously monitor AI systems to prevent abuse and ensure data security
Samir stresses that without these measures, uncontrolled use of AI can not only compromise consumer confidence, but also expose organizations to severe legal and reputational repercussions.
Fake News's and ethical challenges in AI
The spread of fake news AI-generated is another growing concern. “O combating AI-generated fake news requires a combination of technology and education. Automated verification tools, identification of synthetic patterns in images and texts, and labeling of AI-generated content are important steps we need to invest in public awareness, teaching to identify reliable sources and question dubious content” states Samir.
Ensuring transparency and ethics in AI development is one of the pillars advocated by Samir.He points out that “some of the best practices include adopting explainable models (XAI 5 EXPLAINable AI), independent audits, using diverse data to avoid bias, and setting up AI ethics committees
One of the major cybersecurity concerns associated with AI include sophisticated attacks such as the phishing . An attack technique in which criminals try to trick individuals into revealing sensitive information, such as passwords and banking data, by posing as trusted entities in digital communications. These attacks can be even more sophisticated when combined with AI, creating personalized emails and messages that are difficult to distinguish from the real ones. To mitigate these risks, Samir suggests that “is investing in AI-based detection solutions, implementing multi-factor authentication, and ensuring that AI models are trained to detect and mitigate manipulation attempts is critical
Collaboration for effective AI Policies
Collaboration between businesses, governments and academia is vital to effective AI policymaking.Samir points out that “AI impacts a number of industries, so regulation needs to be built collaboratively. Companies bring the practical view of the use of technology, governments establish security and privacy guidelines, while academia contributes research and methodologies for safer and ethical development
The multifaceted nature of artificial intelligence means that its impacts and applications vary widely across different sectors, from health to education, to finance and public safety.
Enterprises they are fundamental in this process, as they are the ones who implement and use AI at a large scale actionable insights, on market needs, practical challenges and the latest technological innovations.The private sector contribution helps ensure that AI policies are applicable and relevant in the real context.
Governments, in turn, have a responsibility to establish guidelines that protect citizens and ensure ethics in the use of AI. They create regulations that address issues of security, privacy and human rights.In addition, governments can facilitate collaboration between different stakeholders and promote funding programs for AI research.
Academy 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 enables AI policies to be robust and adaptable, addressing both the benefits and risks associated with using 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 respect security and privacy standards.
Samir points out 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 strike a balance between regulation and freedom to innovate so that we can maximize the benefits of AI while minimizing risk,” 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 operation and impact.
To clarify, demystifying these points, and finalizing the interview, Samir Karam answered several questions in a ping-pong format, addressing the most common myths and providing actionable insights, valuable about the reality of AI.
- What are the most common myths about artificial intelligence that you encounter and how do you clarify them?
One of the biggest myths is that AI is infallible and completely unbiased. In reality, it reflects the data it was trained with, and if there are biases in that data, AI can reproduce them. Another common myth is that AI means complete automation, when in fact, many applications are just assistants for decision making.
- AI really can replace all human jobs?What is the reality about it?
AI will not replace all jobs, but it will transform many of them. New roles 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 dominate 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 intention of their own.
- Are all artificial intelligences dangerous or can they be used for harmful purposes? What should we know about it?
Like any technology, AI can be used for good or ill. The danger lies not in AI itself, but in the use that is made of it. That is why 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.In addition, AI models can be easily fooled by adversarial attacks, where minor manipulations in the data can lead to unexpected results.
- Is AI just a fad or is it a technology that is here to stay?
AI is here to stay. Its 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 really capable of making completely unbiased decisions? How can biases affect algorithms?
No AI is completely unbiased. If the data used to train it contains bias, the results will also be biased. Ideally, companies should adopt bias mitigation practices and conduct constant audits.
- All AI applications involve surveillance and personal data collection?What should people 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 legislation such as the LGPD (General Data Protection Law) and the GDPR (General Data Protection Regulation (General Data Protection Regulation) are fundamental.
