Chinese company DeepSeek’s AI model has sent shockwaves within the AI community. DeepSeek’s model demonstrates that AI can deliver affordable solutions. However, the company’s Chinese origin raises important questions surrounding ethics and transparency.
Questions about DeepSeek’s compliance with Indian laws and sensitivity toward territorial issues have already surfaced. When specifically asked about India’s North-Eastern States or Kashmir, the app reportedly responds, “Sorry, that’s beyond my current scope. Let’s talk about something else.”
These responses or lack thereof, are red flags, particularly given India’s contentious history with China. Due to security concerns in the past, India had banned over 300 Chinese apps, including TikTok. Banning precedents on account of territorial sovereignty and data security places DeepSeek in tricky waters and it is to be seen what approach India will take. But India’s approach will also be informed by the recent thaw in relations with China.
India’s legal arsenal
India’s ability to ban applications such as DeepSeek AI is well-supported by its legal framework. The primary mechanism for such actions is Section 69A of the Information Technology Act, 2000 (IT Act). Section 69A empowers the central government to block public access to information.
In addition to Section 69A, Rule 3(1)(d) of the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (IT Rules), provides an additional provision for takedown and there are previous precedents for take down of app under both provisions. While there is some legal uncertainty around the takedown process under Rule 3(1)(d) read with Section 79(3)(b) of the IT Act, in absence of a judicial stay the takedown regime is valid and enforceable.
Supreme Court in Shreya Singhal v. Union of India, (2015) 5 SCC 1 (2015), upheld Section 69A’s constitutionality. In the same judgment, the Court had struck down Section 66A of the IT Act as unconstitutional for being overly broad and vague. While upholding the fires of Section 69A, the Supreme Court emphasised procedural safeguards and transparency. Interestingly, Shreya Singhal also read down Section 79(3)(b), to mean actual knowledge as a court order or as directives from a notified government agency.
Given the above scheme, DeepSeek’s refusal to answer questions about India’s North-Eastern States or Kashmir may fall foul of Indian law. DeepSeek’s AI’s failure to engage with sensitive topics, may be construed as undermining India’s sovereignty or territorial integrity, both being reasonable restrictions in terms of Section 69A and Article 19(2) of the Constitution.
Apart from the above concerns, there are broader ethical questions and concerns regarding DeepSeek’s AI. Its reliance on open-source contributions, though cost-effective, also raises questions about data privacy and its adherence to Indian data protections laws.
OpenAI, a US company, is already facing legal challenges in India with respect to copyright issues, which still have to be adjudicated by Indian Courts. In addition, surely, the same challenges will also apply to DeepSeek.
There are also doubts regarding the data and storage practices of DeepSeek, especially in light of India’s latest data protection laws. DeepSeek collects extensive data from its users, clearly evident from its privacy policy. India’s draft data protection rules, being discussed presently, talk about data localisation. It is to be seen in what form, data localisation will be implemented in India, so DeepSeek may find itself in choppy waters, if its privacy policy is to be believed.
Compliance with India’s data protection law by DeepSeek, is still to be seen, but it’s get-political responses, risks eroding trust of Indians on sensitive issues. India has been reluctant in embracing Chinese technology in the past. Transparency by DeepSeek on various issues is critical and until DeepSeek takes steps to address its issues, Indian users ought to proceed with caution.
India now faces a critical decision: whether to ban DeepSeek AI or adopt a more nuanced approach. A ban now may affect India’s thawing of relations with China, however, non-compliance with upcoming data protection laws will have necessary and logical consequences. Further, a ban may discourage innovation and global collaboration in AI development but data privacy and national security concerns have always weighed heavily with the Indian polity, especially with the Chinese.
Bute the larger question is: how can nations manage AI technologies in a way that safeguards sovereignty and data protection, while fostering global collaboration?
The writer is Partner, Shardul Amarchand Mangaldas & Co