AI ecosystem feedback on MeitY’s Advisory

Seeking clarifications and public consultations

On 1st March, 2024, the Ministry of Electronics and Information Technology (MeitY) issued an advisory to the AI industry, broadly stating due diligence requirements for intermediaries and platforms under the IT Act, 2000 and IT Rules, 2021. Industry bodies, in particular the AI ecosystem, shared their concerns regarding the lack of clarity on the Advisory’s intended audience, processes and standards, and legal implications in case of non-compliance.

Feedback from the industry was gathered through conversations within The Fifth Elephant community with developers, founders, lawyers, and policy analysts. The intention of this report is to document said concerns and recommendations, to be shared with MeitY. It was further discussed that the AI ecosystem must be actively involved in the policymaking and consultation process, to ensure minimum friction and delays in implementation.

Key Concerns

  1. The term “platforms” is not defined and it is not clear what specific type of organizations/individuals falls under this category.
  2. Standards for testing models and software for bias, discrimination, and upholding electoral integrity are not mentioned.
  3. Lack of clarity on what comprises “under-testing” and “unreliable” AI models, software, and algorithms.
  4. Which regulatory body or channel within the Government is to be approached to seek permission for deploying “under-testing” and “unreliable” AI models, software, and algorithms, as well as the application process for seeking permission are not specified.
  5. Standards and processes to measure “the possible and inherent fallibility or unreliability of the output generated” are not specified, thereby making the requirement for labeling vague.
  6. The requirement to seek permission from the Government before deploying AI models would significantly hinder start-ups and small businesses using AI, thereby creating a roadblock and possible Big Tech monopoly. This will also have a negative impact on innovation and entrepreneurship.
  7. Measures to counter the spread of misinformation generated through deepfake technology are increasingly insufficient to curb its virality. Apart from identification of such content, it is also necessary to develop security measures, balancing privacy, free speech, and innovation rights.
  8. There is uncertainty surrounding the legal implications of non-compliance with the Advisory due to unclear legal basis, the lack of publicly available sources for the advisory, and ambiguity regarding the specific sections under which these requirements are placed. It is also unclear whether intermediaries or platforms who have not directly received the Advisory from the Ministry also need to submit the Action Taken-cum-Status Report, and hence, if the Advisory is binding on the whole industry.
  9. It is unclear whether clarifications tweeted by the MoS are legally binding as the same has not been officially announced by the Ministry.

Recommendations

  1. To extend the deadline for compliance pending public consultation and clarification of the Advisory.
  2. To organize continuous public consultation efforts with the startup and developer ecosystem, lawyers, and social scientists to develop an effective AI regulation regime.
  3. To issue official clarification on concerns raised regarding the Advisory and reconsider the compliance clause pending enforceable regulations following a robust consultation process.

The Fifth Elephant is invested in facilitating interactions between MeitY and representatives from the AI ecosystem to improve the state of AI regulations in the Indian tech ecosystem. To share your insights, leave a comment.

Hosted by

Anthill Inside is a forum for conversations about risk mitigation and governance in Artificial Intelligence and Deep Learning. AI developers, researchers, startup founders, ethicists, and AI enthusiasts are encouraged to: more