Last week, Perplexity announced an upgrade to its AI chatbot on X (formerly known as Twitter), introducing a new feature named Ask Perplexity that allows users to generate videos based on specific prompts. The company has not disclosed which large language model (LLM) powers this video generation capability, but it emphasized the incorporation of safeguards to prevent the creation of synthetic videos resembling real-world celebrities and public figures. This feature is currently accessible to all X users who mention the chatbot in their posts or first-level comments.
Ask Perplexity Can Now Generate AI Videos
The official Ask Perplexity account confirmed the new video generation functionality in a post on X. This positions it as the first and only chatbot on the platform equipped with such capabilities, with competitors like Grok, developed by Elon Musk’s xAI team, lacking this feature.
— Ask Perplexity (@AskPerplexity) June 19, 2025
Ask Perplexity can generate videos lasting up to eight seconds, based on a text prompt alone or a combination of an image input with a text prompt. The generated outputs can include native audio, and users can specify dialogues for multiple speakers. Although Perplexity did not reveal which AI model it employs for this feature, it appears to be utilizing Google’s Veo 3 model based on its described capabilities, though this remains unverified.
Following the announcement, users on X began experimenting with the video generation feature, submitting a range of creative and humorous prompts. Staff members at Gadgets 360 observed several instances where users requested videos featuring real celebrities or public figures. However, the videos produced typically featured characters that bore only a vague resemblance to the actual individuals. Given that users can submit images as prompts, it remains unclear whether Perplexity will accommodate such requests in the future.
— Ask Perplexity (@AskPerplexity) June 19, 2025
The following day, the chatbot responded to user video requests with a humorous tweet, saying, “I’ve read through your video request DMs. Some of y’all need help. Not AI therapist help. Real help.” This lighthearted remark suggests that some of the requests may be too unconventional to fulfill, hinting at the limits of the feature.
