Aravind Srinivas, the Indian-American co-founder and CEO of AI chatbot Perplexity, has found a way to bypass Chinese censorship and address privacy concerns surrounding DeepSeek R1, the viral language model that’s taking the world of artificial intelligence by storm lately.
Perplexity is basically tapping into the “open-source” nature of DeepSeek’s viral language model that allows anybody—individual, organisations, you name it—to deploy and build their own spin-off experiences out of it, free of cost.
And so, by taking the build materials from DeepSeek, Perplexity has been able to fashion its own “R1” for its users and potentially solved one of its biggest loophole—security. Whether or not, the “treatment” makes it anymore secure than any other AI chatbot on the Internet, is still up for debate but at least, Perplexity has been able to pacify concerns that China might someday peek into your computer/smartphone on the back of one of these chatbot, or at least, that is what the thought process is.
DeepSeek R1, an open-source AI model developed in China by an unlikely contender spun off from a hedge fund based out of Hangzhou, has turned the world of AI upside down with its impressive capabilities rivaling those of giants like ChatGPT, wiping off over $500 billion from Nvidia’s market cap. However, as usage grew by leaps and bounds, so have conversations about privacy and data security.
Srinivas and his team at Perplexity have tackled these issues by leveraging the open-source nature of DeepSeek R1. While not explicitly mentioned, they have implemented modifications that ensure uncensored results and secure user interactions. All data is claimed to be stored on US-based servers, at least for US users.
To demonstrate the effectiveness of their solution, Srinivas conducted a live demonstration using the infamous “Tiananmen Square” test case. The results were clear: Perplexity, powered by DeepSeek R1, delivered uncensored information without any interference proving Perplexity’s integration of DeepSeek R1 circumvents censorship to access seemingly unbiased information.
To access the full benefits of DeepSeek R1, users can subscribe to Perplexity Pro for $20 per month or $200 per year. Perplexity is also aggressively ramping up on the amount of queries users have access to increasing it from 100 to 500 within the span of a few hours. This subscription also unlocks access to other powerful AI models, including OpenAI’s GPT-4. Free users get five daily queries, Srinivas has announced.
Follow FE Tech Bytes on Twitter, Instagram, LinkedIn, Facebook