On May 7, 2025, Pinterest announced a series of enhancements aimed at improving its visual search experience, leveraging advanced artificial intelligence (AI) technologies. This initiative is part of Pinterest's effort to position itself as a leading visual search engine, enabling users to discover aesthetic elements that are often hard to describe in words, such as styles, atmospheres, and trends.
Among the most significant updates is the introduction of Visual Language Models (VLM), a form of generative AI designed to help users better understand their visual preferences. Initially rolled out in the women’s fashion section in the United States, Canada, and the United Kingdom, this feature allows users to tap on a Pin and see an animation that highlights the most descriptive words related to that image. According to Pinterest, this technology will automatically generate terms that help identify key aspects of an image, including color, fabric, or style.
Moreover, Pinterest aims to refine fashion searches further by allowing users to seek alternatives to items they find appealing but with specific variations, such as color, style, or fabric. For instance, if a user discovers a blazer they like, the platform can suggest more formal options suitable for elegant events or alternatives featuring different details. This advanced visual search capability will make online shopping experiences more personalized and targeted, catering to users looking to explore similar styles that are more appropriate for specific occasions.
Additionally, Pinterest plans to extend this functionality across more sections of its app, enabling users to perform visual searches directly from the home screen by simply pressing and holding on a Pin. This enhancement will provide immediate access to a broader range of visual content, improving both the speed and accuracy of searches.
In response to user concerns regarding the quality of content and the authenticity of images, Pinterest has also decided to label images generated by AI. This update addresses worries that AI could compromise the quality of content and diminish the discovery of genuine images. With the introduction of a clear label, users can easily distinguish between real images and those created by algorithms, allowing them to limit the number of AI-generated images displayed in their search results. This choice stems from numerous critiques received from users who were apprehensive that the increasing use of AI could undermine the visual experience on Pinterest.
Meanwhile, Axis Communications, a leader in network video surveillance, released its latest research titled 'The State of AI in Video Surveillance' on the same day, exploring the implications of AI in the realms of security, protection, business intelligence, and operational efficiency. The study provides an overview of critical aspects of AI technologies, their integration into existing systems, and the significant opportunities and challenges that arise from their application.
The implementation of AI has surged notably in the past two years, driven by a marked increase in market demand and an improved understanding of applications, alongside the emergence of new use cases. The research identifies AI and generative AI, alongside cybersecurity, risks and privacy, and analytics, as the most significant trends shaping the future of video surveillance and operations.
Notably, 34% of end clients believe that AI and generative AI are genuine game changers for their businesses. This sentiment almost doubles among channel experts, with 62% classifying these technologies as significant trends, and 66% of top management indicating them as the primary trend within the industry.
Furthermore, the research highlights several key elements influencing the sector, including the acceleration of cloud and edge AI transition. The findings reveal that the shift from on-premise server systems to hybrid architectures is ongoing, driven by the need for greater scalability, faster processing, and improved bandwidth utilization. The hybrid model, which combines the immediate processing capabilities of edge AI with the long-term data storage and scalability of cloud solutions, is emerging as the preferred approach for many organizations.
Another critical insight from the research is the integration of additional sensory data, such as audio and environmental context, which will enhance scene evaluation and provide more actionable insights. This comprehensive approach is expected to revolutionize security and protection measures while simultaneously elevating business intelligence capabilities. The combination of diverse data streams allows for more precise detection and prediction of potential threats. For example, in emergency scenarios, the pairing of visual data with audio analysis enhances the speed and accuracy of response from security teams.
The study also notes the increasing adoption of facial recognition technology across various countries, bolstered by new regulations that clarify how this technology can be ethically applied within the video surveillance sector. According to Axis experts, facial recognition is set to gain further traction globally, but it is crucial to align its use with privacy regulations and ensure greater transparency regarding the technology's functioning.
Ethical considerations surrounding the use of AI remain a central focus, particularly in regions with stringent privacy laws. The research emphasizes that the responsible use of AI and ethical considerations are viewed as fundamental priorities in the development and deployment of AI systems. Many interviewees expressed concerns that AI-driven decisions could be biased or unreliable, raising potential privacy violations related to facial recognition and behavior analysis.
Moreover, the importance of incorporating responsible AI practices from the early stages of development is highlighted, advocating for the design of surveillance systems that prioritize fairness, transparency, and data privacy from the outset rather than attempting to integrate these aspects later on. While the introduction of clear regulations has been widely recognized as necessary for building trust and accountability, several respondents also underscored the need for a balanced approach that safeguards innovation while addressing privacy and data security challenges.
As Pinterest and Axis Communications unveil their advancements in AI, it is evident that the integration of these technologies will significantly influence user experiences and operational efficiencies across various sectors, from visual search to video surveillance.