Last month, OpenAI’s Codex possibly solved the Navier-Stokes equations in days. But mathematician Tristan Buckmaster feared that his unpublished drafts had fed the model. So, why has the episode sparked concerns related to credit and ownership in the AI era?
The Buckmaster Saga: At the center of the dispute are the century-old equations, which carry a $1 Mn prize. After Buckmaster uploaded his unpublished research into the unreleased model, the AI giant announced that it had solved a key proof in four days. Though OpenAI denied using user inputs for that specific output, the episode exposed how feeding confidential academic work into AI tools could erode priority and career-defining credit.
India’s AI Rulebook: Amid the row, Indian academia is moving from instinct to policy. Many local researchers use AI only for literature review, but avoid uploading complete solutions or patentable methods. Institutes are also running smaller open-weight models on campus servers, discouraging frontier models for sensitive data, citing lack of user control.
Many colleges are also framing guidelines for responsible AI use, including secure handling of unpublished research.
The Policy Gap: India does not have a standalone law that governs AI in research. Even DPIIT’s proposed 2025 framework focuses primarily on financial compensation. Legal experts argue that this fits creators like musicians and not scientists, where harm is also loss of credit and not just revenue. Without explicit protections, scientists who input draft manuscripts or unfiled patents onto AI servers could face loss of proprietary rights.
Safeguarding Research: Experts recommend a strict protocol for scientists: file provisional patents before using AI, utilise APIs with guaranteed non-training clauses, and maintain dated offline notebooks to prove invention timelines. As such, IP attorneys believe that researchers must self-police what they share with AI until clearer rules emerge.
As usage of AI for research surges and institutions formalise ethical AI guidelines, how can researchers safely use AI without exposing unpublished work? Let’s find out…
Analogue chip design still depends on thousands of simulation runs and manual parameter adjustments. This makes the process slow, expensive and difficult to scale. circuitEvolve is automating this repetitive workflow with its in-house AI layer.
AI For Chip Design: Founded in 2026, circuitEvolve is building AI-powered solutions that work alongside existing electronic design automation software. Once engineers input target specifications, the platform then searches through device configurations to find designs that balance performance, power and area.
The Closed Loop Stack: The startup’s flagship product, OPTIM, simulates circuit parameters, evaluates each candidate and verifies the results against target constraints. It continues iterating until the design converges. The system is simulator-agnostic and works across process-design kits and device libraries. The startup claims that OPTIM has been validated across more than 1,000 devices.
Riding India’s Chip Wave: The startup’s broader ambition is to automate the analogue design workflow from specification to layout-ready output. Future applications could include circuit understanding, schematic generation, simulation analysis and AI-guided design.
With the homegrown semiconductor market projected to become a $200 Bn opportunity by 2035, can circuitEvolve make AI an everyday co-pilot for analogue chip engineers?
Indian startups raised $2.2 Bn+ in Q3 2026. But the biggest takeaway was that AI and cleantech emerged as the most-funded sectors during the quarter, piping industry favourites fintech and ecommerce. Here is all about it…
Source: Inc42
