Google has officially launched Veo, its latest generative AI video model, enabling businesses to incorporate advanced video creation into their content generation processes. The unveiling comes after a private preview via Google's Vertex AI platform, which started in May, that allows users to explore the technology ahead of its broader release.

Veo is designed to generate high-quality videos at 1080p resolution, offering a variety of visual and cinematic styles based on either text or image prompts. Initial demonstrations of the model indicated that the generated videos could exceed a minute in length; however, Google has not specified any length limitations for this preview release. Recent example clips provided by Google showcase the model's capabilities, illustrating that the video's AI-generated nature is often indiscernible to viewers.

An instance highlighted within the announcement showcased a dog animated with consistent fur patterns and collar details throughout its movement, underscoring the model's sophistication. However, it is flagged that the technology is not without flaws, as evidenced by certain clips exhibiting issues like unnatural lighting effects, including a notable instance where light shines through a person's hand.

In addition to Veo, Google is set to expand access to its Imagen 3 text-to-image generator for all Google Cloud customers beginning next week. This follows an initial release in the United States in August through Google’s AI Test Kitchen. Users included in Google’s allow list will gain access to enhanced features, such as prompt-based photo editing and the ability to customise generated images with brand elements like logos and product features.

To address concerns surrounding the generation of harmful content and copyright violations, both Veo and Imagen 3 are equipped with built-in safeguards. However, there remain vulnerabilities in the copyright protection measures, as users have been able to bypass these restrictions. Furthermore, to combat issues related to misinformation and content misattribution, all materials produced by Veo and Imagen 3 are embedded with DeepMind’s SynthID technology. This invisible digital watermark functions similarly to Adobe's Content Credentials, enabling transparent identification of AI-generated content.

As these technologies proliferate, the potential for AI automation in business practices continues to evolve, with implications for content creation across various sectors.

Source: Noah Wire Services