Tietê Bus Station in São Paulo establishes exclusive area for Uber passenger pickup

Since Thursday, over 5000 pilots from Air India and Air India Express have begun undergoing 'mandatory substance screening'. The airline has taken this step with the aim of 'protecting and maintaining' passenger trust.

Passenger trust suffered after the accident in Ahmedabad last June and was slowly recovering, but it was shaken again after the crew commander of a flight from Phuket to Delhi on August 4 failed a drug test.

Following this large-scale one-time check, AI will conduct tests in accordance with regulatory requirements, which currently mandate testing at least 10% of the crew annually. The DGCA is reviewing its rules and may introduce stricter consequences for those who test positive due to addiction issues, as well as adopt a different approach to those who test positive after taking medication for pain, allergies, and other ailments.

The Senior Vice President of AI for Flight Operations, Captain Manish Uppal, wrote to all pilots on Thursday, emphasizing that AI has 'earned the trust of millions of passengers and the general public' over decades: '...This trust is our most important asset, and it is vital that we do everything possible to protect and preserve it.'

OpenAI, the developer of ChatGPT, has achieved an annualized revenue exceeding US$40 billion, which is equivalent to about R$207.7 billion, as reported by Bloomberg on Thursday, the 13th. This amount represents approximately double the revenue pace the company showed at the end of 2025.

The term used, known as 'run rate', is a projection that estimates the company's revenue generation rate over a twelve-month period. It is important to note that this figure does not confirm that OpenAI has already generated US$40 billion during the year.

This acceleration in results occurs while the company expands its artificial intelligence (AI) services and prepares for a potential public stock offering. Competitor Anthropic has also shown robust growth in the adoption of its tools by corporations, reaching a 'run rate' of US$47 billion (approximately R$244.1 billion) in May, according to data cited by Bloomberg.

However, a direct comparison between the two companies must be made with caution, as both are private entities and may use different methodologies to calculate and disclose their annualized revenues. Nevertheless, the figures highlight the high competitiveness in the corporate AI sector, where technology companies compete for large contracts for the use of their models and resources.

The financial advance happens at a time when OpenAI is approaching a possible Initial Public Offering (IPO). In this scenario, the ability to convert the vast popularity of ChatGPT into continuous revenue is a crucial indicator for investors to evaluate the business. The growth also signals the expansion of the market for specialized AI solutions, going beyond general public chatbots, with companies like OpenAI developing specific tools for professional tasks, such as programming and productivity.

With annualized revenue now surpassing US$40 billion, OpenAI strengthens its commercial dimension while dealing with increasingly intense competition for enterprise clients and participation in the global AI market. The original article was published in Olhar Digital.

WhatsApp is implementing a new artificial intelligence (AI) feature that operates directly on the mobile device to identify potential fraud during conversations. This system, called Scam Alert, examines messages locally and can issue a warning if the model detects characteristics of a scam attempt.

Currently, the functionality is available in a restricted beta version and is optional. According to Meta, the company responsible for WhatsApp, the purpose is to help users recognize suspicious approaches before giving out information, sending money, or taking any other actions requested by criminals.

A crucial aspect of this system is its local processing. This implies that message content does not need to be transmitted to Meta's servers for classification to occur. Meta assures that with the feature active, no message data leaves the user's device for analysis purposes.

Additionally, the company reports that no messages are automatically sent for reporting to WhatsApp, Meta, or third parties, and the user retains the freedom to disable Scam Alert at any time. This approach aims to protect communication privacy while employing machine learning models to map patterns related to scams.

Meta's Context and Concerns

Meta's concern is linked to the vast scale of WhatsApp, which exceeds three billion users globally, making it a channel frequently exploited by criminals to contact potential victims.

Types of scams include fraudulent requests for bank transfers and schemes known as 'pig butchering,' where scammers establish a bond of trust with the victim before persuading them to invest in fake opportunities or send funds.

Data provided by the U.S. Federal Trade Commission (FTC) indicates that victims reported losses of US$ 425 million (equivalent to about R$ 2.3 billion) in WhatsApp-based scams in 2025. This amount represents a fraction of the more than US$ 2.1 billion (approximately R$ 11.5 billion) lost in social media fraud during the same period.

Scam Alert is another measure adopted by Meta to prevent user losses before scams materialize. At the beginning of 2026, the company had already introduced fraud detection mechanisms related to device linking requests on WhatsApp, aiming to prevent criminals from connecting accounts to unauthorized devices.

With this, the company is now expanding the use of AI to internally analyze message content, looking for signs that may indicate a fraud attempt. However, Scam Alert is still in the testing phase, with limited availability, awaiting a full performance evaluation of the system by Meta before being expanded to more users.