Bharat in Betastartup field guide
← Back to the atlas
No. 141ProposalSecurity and governance

Legal AI

Source

The idea

Automate contract analysis and legal research, and build systems that help predict case outcomes.

Why this is interesting

  • Contract review and legal research serve different users: in-house legal teams reviewing agreements, and lawyers or paralegals researching precedent. The likely buyer is a company’s legal function or a law firm. Case-outcome prediction is a separate and more sensitive claim.
  • The source names the functions but not the buyer, jurisdiction or language. Contract analysis is a globally portable software opportunity; research and prediction depend heavily on each jurisdiction’s records and practice, so an Indian version would need its own data and validation. The source provides no India-specific rationale here.

Possible starting points

01

In-house contract review queue

Help a company’s legal team triage routine agreements against its own playbook and route exceptions to a lawyer.

02

Jurisdiction-specific research assistant

Support research within one court system or area of law where the underlying records are accessible and citations can be verified.

Questions to answer

  1. Which legal task is repetitive enough to automate yet low-risk enough that an error is caught before it matters?
  2. Is reliable, lawfully usable source data available for the jurisdiction and practice area chosen?
No. 1192Problem statementSecurity and governance

Improving open source software security using Fuzzing

Develop and demonstrate a fuzz-testing harness for the Windows edition of Sumatra PDF Reader, version 3.5.2 or later, to assess robustness and security. Teams choose their fuzzer, with WinAFL preferred. Submit a working harness and a technical report documenting analysis steps, selected target functions and dependencies, alongside test coverage and findings. This describes authorised software-testing scope without exploitation instructions.

No. 1370Problem statementSecurity and governance

Quantum-Inspired Cyber Threat Detection for Digital Signature Security

Develop a simulated threat-detection and verification framework for teleportation-based quantum digital signatures, explicitly without AI or machine learning. Use quantum-state/measurement principles and statistical thresholds to examine forgery, impersonation, replay and unauthorised verification. Include mathematical modelling, controlled attack simulations, security analysis and performance evaluation. Strong security and low complexity are properties to test, not guaranteed outcomes; the source leaves its deliverables table unfinished.

No. 1421Problem statementSecurity and governance

Trustworthy Computer Vision Integrity Assurance for Data, Models and Inference Outputs in Multi-Contributor Pipelines

Develop an offline, model-agnostic vision-assurance framework assessing contributed training data, model behaviour and inference records. Identify suspicious samples/contributor patterns, evaluate model integrity within declared access limits and cryptographically bind outputs to inputs, weights and processing configuration. Distinguish distribution shift from manipulation where evidence permits, reporting explanations, confidence, actions and tamper-evident history. Support COCO/YOLO and organiser ONNX/PyTorch formats; baseline checks must not require retraining, and white-box methods must degrade honestly without access. Demonstrate reproducible public/team-generated tests and declare coverage gaps.