À medida que a IA avança, todos nós temos um papel a desempenhar para desbloquear o impacto positivo da IA para organizações e comunidades em todo o mundo. That is why we are focused on helping customers use and create AIreliable, that is, an AIsafe, protectedandprivate.  

Na Microsoft, temos compromissos em garantir uma IA confiável e estamos criando tecnologia de suporte de ponta. Nossos compromissos e recursos andam de mãos dadas para garantir que nossos clientes e desenvolvedores estejam protegidos em todas as camadas.

Building on our commitments, we are today announcing newresourcesof products to strengthen the safety, security and privacy of AI systems.

Security.Security is our top priority at Microsoft, and ourSecure Future Initiative(SFI) underscores our company-wide commitments and the responsibility we feel to make our customers moreinsurance. This week, we announced our firstSFI Progress Report, destacando atualizações que abrangem cultura, governança, tecnologia e operações. Isso cumpre nossa promessa de priorizar a segurança acima de tudo e é guiado por três princípios: seguro por design, seguro por padrão e operações seguras. Além de nossas ofertas primárias, Microsoft Defender e Purview, nossos serviços de IA possuem controles de segurança básicos, como funções internas para ajudar a evitar injeções imediatas e violações de direitos autorais. Com base neles, hoje estamos anunciando dois novos recursos: Além de nossas ofertas primárias, Microsoft Defender e Purview, nossos serviços de IA possuem controles de segurança básicos, como funções internas para ajudar a evitar injeções imediatas e violações de direitos autorais. Com base neles, hoje estamos anunciando dois novos recursos:

• Reviews no Azure AI Studio to support proactive risk analysis.
• Microsoft 365 Copilot will providetransparency in web queries para ajudar administradores e usuários a entender melhor como a pesquisa na Web aprimora a resposta do Copilot. Disponível em breve.

Nossos recursos de segurança já estão sendo usados pelos clientes.A Cummins, a 105-year-old company known for its engine manufacturing and development of clean energy technologies,turned to Microsoft Purview para fortalecer sua segurança e governança de dados, automatizando a classificação, marcação e rotulagem de dados.A EPAM Systems, a software engineering and business consulting company, implemented  o Microsoft 365 Copilot para 300 usuários por causa da proteção de dados que eles recebem da Microsoft. J.T. Sodano, Diretor Sênior de TI, compartilhou que “estávamos muito mais confiantes com o Copilot para Microsoft 365, em comparação com outros LLMs (grandes modelos de linguagem), porque sabemos que as mesmas políticas de proteção de dados e informações que configuramos no Microsoft Purview se aplicam ao Copilot.

Security.Including security and privacy, the broader principles ofMicrosoft Responsible AI, estabelecidos em 2018, continuam a orientar como criamos e implantamos IA com segurança em toda a empresa. Na prática, isso significa construir, testar e monitorar adequadamente sistemas para evitar comportamentos indesejáveis, como conteúdo prejudicial, viés, uso indevido e outros riscos não intencionais. Ao longo dos anos, fizemos investimentos significativos na construção da estrutura de governança, políticas, ferramentas e processos necessários para defender esses princípios e construir e implantar a IA com segurança. Na Microsoft, estamos comprometidos em compartilhar nossos aprendizados nesta jornada de defesa de nossos princípios de IA Responsável com nossos clientes. Usamos nossas próprias práticas recomendadas e aprendizados para fornecer às pessoas e organizações recursos e ferramentas para criar aplicativos de IA que compartilham os mesmos altos padrões pelos quais nos esforçamos.

Today, we’re sharing new capabilities to help customers realize the benefits of AI while mitigating the risks:

• A resourceof correction na funcionalidade de detecção de grounding do Microsoft Azure AI Content Safety que ajuda a corrigir problemas de alucinação em tempo real antes que os usuários os vejam.   
• Embedded Content Security, que permite que os clientes insiram o Azure AI Content Safety em dispositivos. Isso é importante para cenários no dispositivo em que a conectividade de nuvem pode ser intermitente ou indisponível.   
• New reviews no Azure AI Studio para ajudar os clientes a avaliar a qualidade e a relevância das saídas e a frequência com que seu aplicativo de IA gera material protegido.  
• Detection of protected material for programming agora está em versão prévia no Azure AI Content Safety para ajudar a detectar conteúdo e código preexistentes. Esse recurso ajuda os desenvolvedores a explorar o código-fonte público em repositórios do GitHub, promovendo colaboração e transparência, ao mesmo tempo em que permite decisões de codificação mais informadas.

É incrível ver como os clientes de todos os setores já estão usando as soluções da Microsoft para criar aplicativos de IA mais seguros e confiáveis. Por exemplo,  a Unity, uma plataforma para jogos 3D, usou o Microsoft Azure OpenAI Service para criar o Muse Chat, um assistente de IA que facilita o desenvolvimento de jogos. Muse Chat uses content filtering models inAzure AI Content Safety para garantir o uso responsável do software. Furthermore, theASOS,a UK-based fashion retailer with nearly 900 brand partners, used the same built-in content filters in Azure AI Content Safety to support high-quality interactions through an AI application that helps customers find new looks.

Também estamos vendo o impacto no espaço educacional.New York City Public Schools fizeram parceria com a Microsoft para desenvolver um sistema de bate-papo seguro e apropriado para o contexto educacional, que agora estão testando nas escolas.The South Australian Department of Educationalso brought generative AI to the classroom with EdChat, relying on the same infrastructure to ensure safe use for students and teachers.

Privacidade.Data is at the heart of AI, and Microsoft's priority is to help ensure that customer data is protected and compliant through ourprivacy principles de longa data, que incluem controle do usuário, transparência e proteções legais e regulatórias. Para desenvolver isso, hoje estamos anunciando:

• Confidential inferencein preview versionem nosso modelo Azure OpenAI Service Whisper, para que os clientes possam desenvolver aplicativos de IA generativa que dão suporte à privacidade verificável de ponta a ponta.The confidential inference garante que os dados confidenciais do cliente permaneçam seguros e privados durante o processo de inferência, que é quando um modelo de IA treinado faz previsões ou decisões com base em novos dados. Isso é especialmente importante para setores altamente regulamentados, como saúde, serviços financeiros, varejo, manufatura e energia.   
• General availability of Azure Confidential VMscom GPUs NVIDIA H100 Tensor Core, que permitem que os clientes protejam dados diretamente na GPU. Isso se baseia em nossas soluções de computação confidencial, que garantem que os dados do cliente permaneçam criptografados e protegidos em um ambiente seguro para que ninguém tenha acesso às informações ou ao sistema sem permissão.
• Azure OpenAI Data Zones para a União Europeia e os Estados Unidos estão chegando em breve e se baseiam na residência de dados existente fornecida pelo Azure OpenAI Service, facilitando o gerenciamento do processamento e armazenamento de dados de aplicativos de IA generativa. Essa nova funcionalidade oferece aos clientes a flexibilidade de dimensionar aplicativos de IA generativa em todas as regiões do Azure dentro de uma geografia, ao mesmo tempo em que lhes dá o controle do processamento e armazenamento de dados dentro da UE ou dos EUA.

We have seen growing customer interest in confidential computing and enthusiasm for confidential GPUs, including from application security providerF5, que está usando VMs confidenciais do Azure com GPUs NVIDIA H100 Tensor Core para criar soluções avançadas de segurança baseadas em IA, garantindo a confidencialidade dos dados que seus modelos estão analisando. And the multinational banking corporationRoyal Bank of Canada (RBC) integrou a computação confidencial do Azure em sua própria plataforma para analisar dados criptografados, preservando a privacidade do cliente. Com a disponibilidade geral de VMs confidenciais do Azure com GPUs NVIDIA H100 Tensor Core, o RBC agora pode usar essas ferramentas avançadas de IA para trabalhar com mais eficiência e desenvolver modelos de IA mais poderosos.

Achieve more with trusted AI

Todos nós precisamos e esperamos por uma IA em que possamos confiar. Vimos o que é possível quando as pessoas são capacitadas para usar a IA de maneira confiável, desde enrich employee experiencesandremodel business processesuntilreinventing customer engagementandreimagine our livescotidianas. Withnew featuresque melhoram a segurança, a proteção e a privacidade, continuamos a permitir que os clientes usem e criem soluções de IA confiáveis que ajudam todas as pessoas e organizações do planeta a alcançar mais.  Em última análise, a IA confiável abrange tudo o que fazemos na Microsoft e é essencial para nossa missão, pois trabalhamos para expandir oportunidades, ganhar confiança, proteger direitos fundamentais e promover a sustentabilidade em tudo o que fazemos.

Imagine receiving a perfect voice message from a celebrity asking for financial help for a social cause or finding a website identical to a renowned store offering unbeatable deals. These are just some of the sophisticated tactics that criminals are adopting through artificial intelligence (AI) to deceive millions of Brazilians daily.

According to a survey conducted on the BrandMonitor platform between March and May of this year, it was identified that 529 out of 628 tracked brands suffered domain attacks and in 405 cases, fake websites emerged applying scams. The 50 most impacted brands registered an average of 86 fake domains created within just 90 days. The technology that promises to transform the future is now being used to sophisticate frauds, creating a digital epidemic that puts consumers and companies on high alert.

One of the main reasons for the growth of AI-driven digital fraud is its versatility, with no limits to the sectors and uses that can adopt it. Despite the versatility of digital attacks, all follow common steps:

1. Data collection:Criminals obtain personal and financial data from consumers and businesses through breaches, phishing and malware.
2. AI Training:With the data collected, AI is trained to identify patterns of behavior and imitate human language convincingly.
3. Execution:AI automates the scam, creating fake profiles, sending personalized messages, and simulating online activities to trick victims.

According to the 2023 global survey on Omnichannel Digital Fraud Trends conducted by TransUnion, attempts at digital fraud have increased by 80% worldwide since 2019. When combined with AI, these frauds become even more sophisticated and convincing to users. In Brazil, the first half of 2024 recorded 800,000 scam attempts.

Examples of scams in Brazil

No Free MarketFor example, criminals create fake profiles with photos and positive reviews generated by bots, misleading buyers into making fraudulent transactions. Already in ride-hailing companies, such as theUberand99 Taxi, AI is used to simulate phantom trips, generating undue charges for customers and illicit profits for scammers.

Although the possibilities are varied, some formats of digital scams are more recognized by users, including:

1. Personalizing Phishing MessagesAI analyzes data about the victim, such as purchase history, interests, and online behavior, to create highly targeted messages. This customization makes it difficult to identify frauds, as the messages appear legitimate and relevant to the recipient;
2. Fake Virtual Assistants:Fake chatbots induce people to provide sensitive information, such as bank details or passwords. Artificial intelligence allows these assistants to respond naturally and adapt to users' questions, making them more convincing;
3. Deepfakes and voice manipulationAI is used to create deepfakes, which are fake videos or audios that appear authentic. This technology is a growing concern, especially in contexts like elections, where the spread of false information can have significant impacts. In February of this year, the Superior Electoral Court decided to restrict the use of AI in campaigns.

Information and education are essential for protection against digital fraud. Users need to be more aware of how these fraudsters operate so that they can have greater control over their decisions, thereby avoiding the scam and all the trouble that comes with it.

Furthermore, to better understand the scope of AI-driven digital frauds, it is crucial to analyze concrete cases that demonstrate the impacts of these practices. Some emblematic examples involving large companies in Brazil reveal how these organizations were targeted by sophisticated scams, exploiting vulnerabilities of both consumers and the brands themselves:

1. Free Market:Criminals use AI to create fake profiles with photos and positive reviews generated by bots, misleading buyers into transacting with non-existent sellers. Additionally, they use AI to create fake product ads, attracting buyers and collecting personal and financial data.
2. Vans:Criminals create fake websites selling Vans products at unrealistic discounts, using AI to personalize these sites and deceive consumers.
3. Fleury:Phishing scams using AI tricked Fleury Group employees, resulting in the leak of sensitive data.
4. ENEM 2024:A fake website imitating the Enem Participant Page led several users to pay the registration fee for the exam, estimated at 85 reais per person.
5. Starlink:SpaceX's satellite internet company, owned by Elon Musk, is constantly victimized by scams and fraud. In our daily monitoring ranking, we found 523 suspicious domains related to Starlink, 152 of which are active. About 44% of complaints about the company are related to fake websites.

AI is also a defense tool

Just as artificial intelligence can be used for scams, it is also a powerful ally in fighting them. It is essential to recognize the importance not only of possessing advanced technology but also of knowing how to apply it effectively through sophisticated pattern detection techniques and behavior analysis.

The digital fraud epidemic in Brazil, driven by artificial intelligence, is a growing challenge that requires constant attention and coordinated action among companies, the government, and consumers. Only with information, education, and the intelligent use of technology can we curb this wave of digital crime and protect our data and our future.

Currently, we live in a time when practicality and time optimization are increasingly valued and demanded. With a busy routine, finding ways to simplify tasks and ensure that basic activities are completed efficiently becomes a differentiator.

Subscription clubs, in turn, emerge as an intelligent solution, offering not only convenience but also the opportunity to personalize the consumption experience. By eliminating the need for frequent shopping and continuous planning, these services provide more time for other important daily demands.

The growth of this sector is significant, as according to data from the Brazilian Association of Electronic Commerce, it increased by 1000% in the last decade in Brazil. Nowadays, there are 4,000 subscription clubs for services and products in operation.

The main advantage is the elimination of concerns about restocking and the significant reduction in time spent on purchasing activities. Instead of setting aside part of the day or week to plan trips to the supermarket, pharmacy, bookstore, or other establishments, the subscriber can rely on regular deliveries, ensuring that essential items are always available.

In addition to automation of refueling, personalization is another fundamental aspect of the model. When signing up for this type of service, the consumer has the opportunity to adjust their preferences, indicating, for example, the type of coffee they prefer most, the size of the clothes they wear, the literary genres they are most interested in, or the dietary restrictions they need to observe.

By receiving only what will truly be used and appreciated, one can avoid the accumulation of unnecessary items, which is especially relevant in a context of environmental concerns and sustainability. The chance to try new brands, flavors, and styles through subscriptions can also broaden the consumer's repertoire, presenting options that might not have been considered in a one-time purchase.

With different proposals, prices, and business models, subscription club services offer alternatives to cater to a wide range of interests and tastes. This means that, regardless of lifestyle or budget, it is possible to find an option that fits each profile.

By simplifying the management of essential items, reducing the time spent on shopping, and offering personalization, they meet the needs of diverse subscribers and reflect an innovative approach to retail and consumption. The sector has become an evolution of the traditional purchasing format, adapting to the demands of a constantly changing society.

The electronic gaming industry is becoming increasingly competitive, as there has been an exponential growth in recent years of active users on online gaming platforms. eSports, as they are known, have increased in popularity due to live broadcasts simultaneously transmitted to various countries, with thousands of competitors worldwide.

This audience is known for beingtech-savvy- English term that refers to people who possess high knowledge and are skilled in the use of technology and digital devices – therefore, they are engaged and loyal to their favorite gaming brands and teams. The diversity of interests and international players makes the eSports audience an attractive and dynamic market.

According to the Game Brasil Research (PGB), 81.2% of respondents revealed that they usually play in the eSports universe, and 88.3% of people watch electronic game matches and tournaments. The time spent on the activity ranges from 1 to 3 hours for 50.8% of the gamers who responded to the survey.

With such expressiveness in the market, companies have recognized the potential that the sector brings to brands and are beginning to use marketing strategies to increase visibility, engage new audiences, and explore new business opportunities.

According to the PGB study, 55.5% of respondents say they use notebooks and computers to play electronic games. Therefore, the commercialization of hardware that improves game performance, as well as technological equipment and accessories for PCs, tends to have many enthusiasts in the gaming market.

High-quality storage devices and memories for PCs, such as internal and external HDDs and SSDs, directly influence the performance of electronic games, providing more dynamism, speed, data processing, and high performance during matches and tournaments. Innovative design cases and keyboards can also be appealing to the gaming audience, who desire more personalized items for the activity.

We cannot forget the importance of having a reliable power supply, such as those from XPG, for example, which ensure a continuous and optimized power flow, greater energy savings, and the security of gaming for hours uninterrupted, without surprises.

In addition to hardware sales strategies, advertising campaigns on gaming platforms, influencer content, and sponsorships are among the marketing options that companies can invest in within this scenario.

Brands can leverage the popularity of eSports by sponsoring events, teams, or streamers. Live stream ads, digital banners, and targeted campaigns within games are also effective ways to reach the electronic gaming audience. Furthermore, influencers have an intimate relationship with their audiences and can promote products and services in an authentic and convincing manner.

An important tip for those planning to do marketing actions in this area is to seek out organizations, teams, or players who are prepared to produce quality content and deliver professional and satisfying results for the client. In this way, it is possible to achieve the much-desired return on investment in the marketing world and establish a lasting and mutually beneficial partnership.

Therefore, entering the eSports market can be challenging due to the need to understand the culture and dynamics of the community. Companies need to invest time in research and establish authentic connections with the audience, whether for PC parts marketing strategies or brand visibility.

*Thiago Tieriis Marketing Manager at ADATA/XPG in Brazil.