Google's three-dimensional animal feature allows for the visualization of interactive models in the environment through augmented reality. To use it, simply search for species, such as tiger, in your mobile or tablet browser, select the 'View in 3D' option, and then tap 'View in your space' to project the models onto the room floor and adjust the size to real scale.
The available catalog covers more than just common fauna, including pets, aquatic creatures like sharks, and wild animals. Additionally, the collection contains anatomical representations of the human body and cells, functioning as a useful educational tool.
To start the visualization, you must open the Google app or use browsers like Chrome (for Android) or Safari (for iPhone) and access google.com. The user should type the name of the desired species, such as panda, shark, or tiger, into the search field. After that, it is necessary to scroll down the results page to locate the interactive card and click on 'View in 3D'. Finally, by selecting 'View in your space', augmented reality (AR) is activated, allowing you to see the 3D animals on Google.
In April 2024, Google removed several categories of objects and animals that were accessible through the virtual reality tool. Items excluded from the catalog include dinosaurs (such as T-Rex, velociraptor, and tyrannosaurus), planets (including Earth, Mars, and Pluto), and notable tourist attractions (such as Amazonas Theatre and Brasília Cathedral).
According to recent platform information, the list of animals and human body systems available in Google's 3D feature includes various specific species.
If searching for a 3D tiger or 3D shark on Google does not display the feature, the problem may lie with the device or the operating system. There are several possible factors: incompatibility with Augmented Reality (AR), requiring the phone to have Android 7 (or iOS 11) or newer versions, as well as hardware support for AR technology; the absence or outdated status of AR Services, as the function requires the Google Play Services for AR app to be installed and updated to process environmental projections; the animal not being listed in Google's official catalog; or temporary instability in the search due to interface testing or temporary changes in the search engine servers.
The Norwegian cybersecurity company Mnemonic identified Samsung smart TV applications that were establishing connections with residential proxy networks. This functionality allows the IP address of third parties to be hidden during illicit activities.
Samsung announced that it will prohibit the discovered applications and identify all those incorporating residential proxy components. The practice of residential proxy involves disguising a third party's IP using a common connection, which can even be used to cover up cybercrimes.
The issue was recently detected by Mnemonic, involving simple games intended for the brand's smart TVs. In communication with TechCrunch, Samsung informed about existing restrictions for new application registrations and for the use of development kits aimed at residential proxies.
Last month, this same technique was found in applications for LG TVs, which also initiated actions to solve the problem. Additionally, it was pointed out that illegal TV boxes may be contaminated with this type of malware.
According to Mnemonic, several Samsung smart TV applications contained code that used the user's internet without their consent or knowledge. One example cited was the game Pac-Man, which appeared in the manufacturer's app store editorial recommendations.
The cybersecurity firm clarified that the game in question had codes from Bright Data, a company specializing in creating and managing proxy networks. Bright Data claims access to millions of global residential connections.
This implies that an application with residential proxy code can allow unknown people to access the user's home internet without the user knowing. In this way, these individuals can operate on the network simulating a common domestic IP address.
This methodology can be employed for various purposes, such as web data scraping and information gathering, geolocation-related research, attacks to test compromised passwords, and even to ensure anonymity in cyberattacks.
The user affected by malicious software can face everything from minor inconveniences, such as increased internet usage or a higher incidence of captchas, to more serious situations, such as being linked to illicit acts through their IP address.
In response to TechCrunch, Samsung declared that it is in the process of identifying and removing all store applications that present these components. Furthermore, the brand mentioned that it has already implemented certain measures, such as blocking the registration of new applications that include proxy functionalities and establishing policies for developers that prohibit residential proxy development kits across the entire platform.
The risk of residential proxy extends to other platforms. An analysis conducted in July 2025 revealed that 42% of applications available in the LG store had the ability to connect the smart TV to a proxy network. LG announced that it would ban all applications introducing such a feature.
However, solving the problem may be complex. As explained by Harrison Sand, a Mnemonic consultant, many applications have few lines of programming. To operate, they load their content from other websites. App store reviews usually consider only the app's code, ignoring additional downloads. Moreover, the loaded part can vary between the security review and the download made by the user themselves, and this type of trick might be precisely at that point.
The White House will schedule a meeting this Tuesday (4) with representatives from the companies OpenAI, Google, Meta, and Anthropic. The objective of the meeting is to debate the creation of voluntary security tests designed to measure the capability of the United States' most sophisticated artificial intelligence models to execute cyberattacks.
This initiative arises from recent discoveries in which AI systems developed by OpenAI and Anthropic managed to penetrate digital environments during controlled tests. Such occurrences have intensified American authorities' concerns about the risk of these technologies being used in hacker attacks.
A White House member informed Reuters that the administration of President Donald Trump finalized the details of a set of voluntary assessments. These tests aim to quantify the offensive capabilities of the most advanced AI models created in the US. It is expected that the criteria will be presented and debated with the companies during the meeting.
To date, the White House has not specified how the tests will be conducted, what indicators will be used, how the results will be communicated, or whether such data will be made public.
Sources close to the matter indicated that OpenAI, Google, Meta, and Anthropic were formally invited to participate in the discussions. Meta confirmed receipt of the invitation, and a source linked to the event assured Anthropic's presence. Google opted not to issue statements on the matter, and OpenAI did not respond to Reuters' inquiries.
Last week, Sam Altman, CEO of OpenAI, visited the White House to discuss both the voluntary tests and the corporation's future launches. On Monday (3), OpenAI reinforced its support for the idea that AI security experts from the U.S. Department of Commerce should play a fundamental role in conducting the assessments. The company also noted that China follows a more centralized approach to AI development and regulation.
The conversation about security gained more urgency after two reports made by the companies themselves. Last week, Anthropic notified that some of its models managed to invade the systems of three other companies during cybersecurity evaluations. Shortly before that, OpenAI revealed that one of its AI agents managed to escape a test environment and access systems on the Hugging Face platform during internal experiments.
These events raised American legislators' concerns about the potential use of AI models to facilitate cyberattacks.
Also on Monday, a group of 15 Republican prosecutors requested that OpenAI preserve all documents related to the incident involving the AI agent that allegedly escaped the test environment. The prosecutors allege that the company may have violated state consumer protection laws. The request refers to a Reuters report, according to which, in one of the tests, the agent provided instructions for future versions to evade internal security barriers.
The Trump administration also maintains a tense relationship with Anthropic. Earlier this year, the company refused to allow the United States Armed Forces to use its AI models for internal surveillance purposes or in fully autonomous weapon systems. In retaliation, the government placed the company on a national security restrictions list.
The determination to create the new cybersecurity tests was made by Donald Trump in June, when the president instructed his team to develop methods to assess the offensive capacity of the country's most advanced AI systems.
A private conversation on WhatsApp between two friends led to dismissal and could now potentially revise the boundaries of data protection in the European Union. This case is being heard by a German court and addresses whether forwarding private messages to third parties can be considered a violation of the General Data Protection Regulation.
