Telangana Tax Officials Barred From Putting Taxpayer Information Into ChatGPT And Other Public AI Tools: Could Employees Face Action For Copy-Pasting Sensitive Data Into AI?

Telangana’s Commercial Taxes Department has drawn a hard line around public AI tools as officers use AI for drafting, translation and case analysis. Circular No. 1/2026, dated 18 August 2026, permits generic research and drafting, but forbids identifiable taxpayer information from being sent to public or commercial AI platforms unless authorised in writing.

Key Highlights

  • Telangana tax staff cannot paste, upload, scan, photograph, dictate or transcribe taxpayer data into unauthorised AI tools.
  • The restriction covers ChatGPT-style chatbots, document analysers, summarisers, translation services and cloud analytical software.
  • Officers must delete previously uploaded taxpayer material, chat histories and stored files where technically feasible.
  • Violations can lead to departmental scrutiny, while certain wilful disclosures may also trigger statutory penalties.

What Exactly Has Telangana Banned From Public AI Tools?

The official Telangana Commercial Taxes Department website now carries the AI circular issued to officers. It lists GSTIN, PAN, legal or trade names, addresses, contact details, bank information, returns, financial statements, invoices, e-way bill data, audit material, investigation records and internal government communications among the protected material.

The wording is broad. An employee cannot escape the rule by uploading a photograph instead of copying text, or by using a personal phone, private email account or cloud storage. The same restriction applies to AI transcription, translation and document-summary services when taxpayer information is involved.

Officers may still use AI to explain a general tax concept, improve generic language or draft non-case-specific formats. Any legal proposition, statutory provision or court citation generated by AI must be checked independently before it enters an official proceeding.

Could Employees Face Action For Copy-Pasting Sensitive Taxpayer Data?

Yes, depending on the facts. The circular makes responsibility personal and says deviations involving confidential taxpayer particulars or orders showing non-application of mind will be viewed seriously. It also cites Telangana civil-service conduct and Classification, Control and Appeal rules, so an internal inquiry or other service action is possible where misconduct is established.

There is also a statutory layer. Section 133 of the Telangana Goods and Services Tax Act says a covered person who wilfully discloses specified information outside authorised duties can face imprisonment of up to six months, a fine of up to ₹25,000, or both. A government servant cannot be prosecuted under that section without previous government sanction.

That does not mean every accidental paste automatically produces a criminal case. The applicable provision, intent, nature of the information, authorisation and departmental process would all affect the outcome.

Why Is Telangana Tightening AI Rules Now?

The circular says tax officers increasingly use AI for discrepancy reports, notices, orders, translation and case analysis. That can save time, but taxpayer files contain financial and identity data that government officers receive through statutory powers.

India has already moved in the same direction elsewhere. In January 2025, the Union Finance Ministry advised employees to avoid ChatGPT, DeepSeek and similar AI applications on office devices because of confidentiality risks. A later Parliamentary answer on government use of AI said officials remain bound by secrecy requirements and should not place classified or sensitive information on insecure platforms or services hosted outside India.

Telangana’s approach goes further by spelling out prohibited input methods, requiring deletion of older AI histories where feasible and making supervisors responsible for checking compliance.

What Should Tax Officers And Taxpayers Watch Next?

For officials, the safest dividing line is simple: public AI may assist with generic work, but identifiable case material should stay inside approved government systems. The circular also requires officers to retain their own judgement in quasi-judicial work rather than copy an AI-generated answer into a notice or order.

For taxpayers, the instruction creates a clearer accountability trail. If confidential GST records appear to have been shared with an unauthorised third-party platform, the issue can potentially raise both departmental and statutory questions. The policy is also likely to influence how other state departments write their own AI-use rules as generative tools become routine in government offices.

FAQs

Can Telangana tax officers use ChatGPT for general research?
Yes, generic research is allowed when no identifiable taxpayer information or case-specific material is shared.

Does the circular cover personal laptops and mobile phones?
Yes, personal devices, private email accounts and cloud storage cannot bypass the taxpayer-data restrictions imposed.

Must officers delete taxpayer data previously entered into AI tools?
Yes, they must delete prior taxpayer data, chat histories and stored files wherever technically feasible.

Can an officer rely directly on an AI-generated court citation?
No, officers must independently verify statutes, judgments and legal propositions before using them in proceedings.

Can every AI-related mistake lead to six months’ imprisonment?
No, criminal liability depends on statutory coverage, wilful disclosure, evidence, sanction and the case’s facts.

India’s Digital Shift: AI, Apple, Safety & Tech

  1. Can Gnani’s AI Challenge Foreign LLMs?
    Check out how Evon 3.3 could strengthen India’s home-grown AI capabilities.
  2. What Changes With Apple’s New CEO?
    Find out what John Ternus replacing Tim Cook could mean for Apple’s future products.
  3. Why Is Roblox Facing Safety Questions?
    Uncover what the new PIL seeks to examine about child safety on Roblox.
  4. How Was ChatGPT Used For Impersonation?
    Reveal how generative AI was allegedly used to create fake PMO documents and gain access.
  5. Will ChatGPT Ads Change Your Experience?
    See who will encounter sponsored content in India and whether it affects AI answers.

Related Articles