Modern organisations generate staggering volumes of electronic data, turning every complex investigation into a race against scale. Investigators who once relied on methodical, document-by-document review now face huge volumes of electronic data and documents far beyond any human team could reasonably examine. That's where AI entered the space, promising speeds that eliminate human limitations.
However, this advantage cuts both ways. The same technology that empowers investigators is now readily available to fraudsters. Threat actors use them to breach secure systems, fabricate evidence, tamper with records and orchestrate fraud at an unprecedented level of sophistication.
Digital Forensics: Speed Comes With A Catch
The initial phases of an investigation look radically different today. AI-enabled forensic tools have quietly absorbed tasks that once consumed weeks of examiner time: sorting and de-duplicating massive evidence pools, recovering deleted files, flagging material transactions by relevance, and spotting behavioural patterns across multiple data sources all at once. This acceleration is undeniable.
Yet, digital forensics is governed by a non-negotiable rule: if you cannot explain it, you cannot use it.
This is where AI faces its most serious challenge. Many AI systems operate as opaque “black boxes” that deliver conclusions without a transparent trail showing how those results were produced. Forensic evidence, especially when presented in legal proceedings, demands a clear and auditable chain of reasoning. An algorithm that flags a file or transaction as suspicious without explaining the underlying risks being excluded from the work product altogether.
Additionally, AI models trained on historical cases and datasets invariably inherit their biases. Beyond operational hurdles lies a deeper structural gap: there is still no universally accepted, court-tested framework, globally or in India, for admitting AI-generated forensic findings as primary evidence.
eDiscovery: AI Is In The Dock Too
The legal fraternity has embraced AI-driven document review at scale, but courts are scrutinising its use with increasing intensity. In 2025, a U.S. court ordered the production of 20 million AI-generated output logs in a copyright dispute, making it clear that AI work-product is not automatically protected. The 2025 “State of AI in eDiscovery” report highlighted a 95% jump in AI enterprise usage in 2024. However, accuracy and reliability have risen to become the second-biggest risk for legal professionals, up from fourth place in 2023.
This has led to a paradox. Legal teams are becoming more cautious and are insisting on stronger oversight and verification before trusting AI-driven document review in legal proceedings. They face a strange new reality: use AI to manage discovery while also expanding human oversight to review and validate AI's decisions.
Cyber Threats And Incident Investigation: The Threat No One Saw Coming
The most significant shift in cyber incident investigation is not a novel attack technique. It is credibility. Attackers now look and sound exactly like the people you trust. A global engineering firm lost $25 million after an employee transferred funds following a video call with senior leadership. Every participant on that call, including the CFO, was an AI-generated deepfake.
Similar incidents, involving AI-cloned voices of senior executives, are being reported more frequently. These attacks arrive through an organisation's trusted internal channels, wearing familiar faces and speaking in familiar voices.
With just a few seconds of recorded audio, attackers can generate convincing voice clones, real-time deepfake videos and precise social-engineering campaigns at a scale previously impossible. Investigators must now question evidence that looks and sounds authentic and behaves authentically.
Adding to the challenge, research shows that AI investigation systems can be manipulated through poisoned training data, creating hidden vulnerabilities that go undetected until the damage is done.
Government In Action
Recognising these risks, governments are beginning to act. At the AI Impact Summit in New Delhi, Prime Minister Narendra Modi unveiled the MANAV vision for AI – Moral & ethical, Accountable & transparent, National sovereignty, Accessible & inclusive, and Valid & legitimate. The framework emphasises responsible, human-centric AI aligned with national priorities, insisting that technology must empower people and drive global progress, while keeping humans firmly in control.
Beyond Boon Or Bane: The Real Test Of AI In Investigations
AI in digital investigation is not a simple choice between benefit and risk. It is a question of maturity. Technology reflects the intent, discipline and governance of those who deploy it; when used as an autopilot, it invites complacency and exposure and when used as an amplifier, augmented by human judgement, ethical guardrails and explainability, it becomes transformational.
The future belongs to investigative teams that understand this distinction and treat AI as a force multiplier under human command.
Sachin Yadav is Partner, Deloitte India, and Shailesh Kand is Director, Deloitte India.
Disclaimer: The views expressed in this article are solely those of the authors and do not necessarily reflect the opinion of NDTV Profit or its affiliates. Readers are advised to conduct their own research or consult a qualified professional before making any investment or business decisions. NDTV Profit does not guarantee the accuracy, completeness, or reliability of the information presented in this article.
Essential Business Intelligence, Continuous LIVE TV, Sharp Market Insights, Practical Personal Finance Advice and Latest Stories — On NDTV Profit.
