Government asks ministries to pause deployment of OpenAI, Anthropic AI models

The Ministry of Electronics and Information Technology (MeitY) has advised government ministries to pause the deployment of AI models from OpenAI and Anthropic for cybersecurity and related functions. This advisory follows meetings where representatives from both companies proposed using their models for government functions. The move aligns with a broader effort by government agencies to restrict the use of unapproved external AI tools for processing sensitive or confidential information to prevent data leaks.

The government's stance on AI has evolved from early warnings about risks to ChatGPT and DeepSeek to the release of the India AI Governance Guidelines, which adopt a risk-based approach. While there is no blanket ban, the government emphasizes using only approved platforms for sensitive tasks. This caution is reinforced by warnings from CERT-In, which noted that advanced AI models could facilitate more sophisticated cyberattacks, necessitating rapid vulnerability patching and strengthened security protocols like multi-factor authentication.

Concerns regarding "shadow AI"—the unauthorized use of AI tools by employees—also drive these restrictions, as uploading sensitive data to public platforms can expose confidential information. Overall, the recent directives indicate a tightening of oversight regarding frontier AI models in official government operations to ensure data security and compliance.

(Source：Medianama)