TechTicker 81: G20 bets on Carolina Principles

We were closely watching the discussions at the G20 Innovation Ministerial last week. With the US holding the G20 presidency this year, frontier tech has emerged as an important part of its agenda — against the backdrop of the US-China technology race and growing debates around the infrastructure needed to power AI.

This month’s Deep Dive looks at what came out of the meeting, including the newly adopted Carolina Principles for Emerging Technologies and what to watch as the US G20 presidency continues.

Closer home, India has proposed new accessibility rules that could change how businesses design and operate digital products for persons with disabilities. In our new Ticker Primer, we break down who the rules apply to, what they require and how businesses can start preparing.

You’ll also get a peek at what’s been keeping us busy at Ikigai Law. And, as always, we have some of our favourite reads from the past month.

I. Deep Dive

Inside the G20 Innovation Ministerial

The G20 is in American hands this year. The US took over the rotating presidency from South Africa in December 2025 and is hosting G20 meetings through 2026, culminating in the Leaders’ Summit end of this year. Technology is emerging as an important part of that agenda. And there is plenty happening in the background.

The US and China are competing for technological leadership, particularly over AI, chips and compute. The rapid build-out of data centres has also become a contentious issue in the US, as communities grapple with their demands on electricity, water and land. These debates formed the backdrop as G20 ministers gathered in North Carolina last week to discuss how governments should approach emerging technologies.

What happened in North Carolina?

On 1–2 September, G20 ministers gathered in Chapel Hill for the G20 Innovation Ministerial.

The meeting brought together G20 governments and some of the biggest names in technology, including Elon Musk, Jensen Huang, Sam Altman, Demis Hassabis, Mark Zuckerberg and Alex Karp. China was represented by Science and Technology Minister Yin Hejun.

The discussions reflected some of the tensions surrounding the AI race. Data centres and the enormous amounts of electricity needed to power AI infrastructure were on the agenda. The US also used the meeting to make the case for a comparatively light-touch approach to emerging-technology regulation: use existing rules where possible and create new ones where genuinely new problems require them.

Despite the differences between G20 members, the meeting ended with a consensus statement.

So, what did everyone agree to?

Six pillars for innovation

The G20 Innovation Ministerial Statement is organised around six areas: a) pro-innovation policy frameworks; b) technology for opportunity and prosperity; c) skilled technical workforces; d) intellectual property and AI; e) AI and technical standards; and f) industrial innovation and supply chains.

There are a few things worth pointing out.

Governments want to use more AI themselves. G20 members agreed to identify and test AI applications across public services, including healthcare, education, transportation and public administration. The statement also recognises what governments will need to do this at scale - better data and digital infrastructure, computing capacity, procurement systems and skilled public-sector workers.

Workers are a big part of the agenda. Members committed to upskilling, reskilling and technical education, alongside an AI Prosperity Compact aimed at encouraging partnerships between governments, companies, universities and research institutions.

The G20 didn't try to solve the AI-copyright debate. The statement recognises both the importance of copyright protection and the difficult questions AI raises for creators and innovators. But on questions such as consent and copyright exceptions, it leaves countries to apply their own laws and legal processes.

And this isn't only about AI. Members also committed to strengthening and diversifying technology supply chains, while identifying issues such as burdensome permitting, involuntary technology-transfer requirements and unnecessary data-localisation mandates as potential barriers to investment and innovation.

But one of the most interesting outcomes of the meeting was a separate document.

Meet the Carolina Principles

G20 ministers adopted the Carolina Principles for Emerging Technologies. These are a common set of principles for how governments can support new technologies as they move from research to the market and, eventually, widespread adoption.

They are not binding, and they are not specific to AI. Each G20 member retains the freedom to develop its own technology policies and regulations.

The Principles cover three stages.

Developing new technologies: Governments should invest in foundational research, bring in private and venture capital, build research and data infrastructure, and make public R&D funding easier to access. They should also build stronger links between researchers, startups, industry and investors to help promising technologies make the jump from the lab to commercial applications.

Getting them to market: The Principles encourage regulatory sandboxes, real-world testing and faster approval processes for emerging technologies. They also give governments another tool i.e. public procurement. Governments can become early customers for new technologies, helping create a market for them and giving companies an opportunity to scale.

Getting them adopted: Governments should first consider whether existing sector-specific rules can govern a new technology. New regulation should be targeted at genuinely new issues that existing frameworks cannot address. The Principles also say governments should consider the opportunity cost of delaying or foregoing the adoption of emerging technologies when deciding how to regulate them.

What to watch out for?

-What happens at home: The Carolina Principles are non-binding. The test will be whether countries pick up ideas such as innovation-focused public procurement, regulatory sandboxes, blended financing and using existing regulation before creating new rules.

For India, some of this is already happening. Questions around getting research out of the lab, using government procurement to create markets for emerging technologies and helping startups cross the gap between R&D and commercialization are already part of India's deeptech-policy conversation. The Carolina Principles now put those questions into a broader G20 framework.

-Watch the US and China: Despite their wider technology rivalry, China signed on to the US-led Carolina Principles. Presidents Trump and Xi are also expected to meet later this month, with AI likely to be on the agenda. What comes out of that meeting will be worth watching.

The G20 Leaders’ Summit will take place at the end of the year, bringing together the leaders of G20 members. The key question for tech policy is whether the priorities agreed in the Innovation Summit, including the Carolina Principles, make their way into the final Leaders’ Declaration

II. From the courtroom to your inbox

Another trend of personality rights cases, but what are the limits: We’ve spent the last few editions tracking India’s growing personality-rights docket. This month, the Delhi High Court continued to protect celebrities against misuse of their identities, while also beginning to ask harder questions about how far these rights should extend.

On 12 August, the Court granted Khushi Kapoor a John Doe injunction against unauthorised use of her name, image and likeness, including through deepfakes, generative AI and voice cloning. A day later, in Aaradhya Bachchan’s case concerning misleading YouTube videos about her health, another bench asked whether the reputation associated with a famous family name extends across generations and whether fake news about an individual can itself amount to an intellectual-property violation.

Then came Janhvi Kapoor. She sought relief against 6,884 URLs allegedly involving everything from pornographic and AI-generated content to impersonation, fan pages and unauthorised commercial use of her identity. The Court ordered 552 prima facie sexually vulgar or pornographic URLs to be taken down but stopped short of granting a blanket injunction over the entire list. It instead asked Kapoor to categorise the remaining material, including by distinguishing direct commercial exploitation from other uses of her name or likeness.

The Court has now appointed advocate Gautam Bhatia as amicus curiae to assist it on the broader legal questions arising from Janhvi Kapoor’s case. A similar exercise is underway in Vivek Oberoi’s personality-rights suit, where advocate Rohan Alva has been appointed amicus.

Courts are continuing to protect public figures against deepfakes, impersonation and commercial exploitation, but are also confronting the other side of the equation: how to do so without creating an expansive right to control criticism, satire, fan content or other legitimate expression. We should be expecting some more definitive orders that address the limits of personality rights.

III. Ticker Primer

India’s new ICT accessibility rules

The Department of Empowerment of Persons with Disabilities (DEPwD) has released draft amendments to India’s disability rules. Any company building, operating or supplying any digital product to Indian users, these rules will likely apply to them. Here is a quick primer about the ICT Accessibility Rules.

Background: Digital accessibility has gained increasing attention through disability-rights advocacy and strategic litigation, particularly as essential services moved online during and after the pandemic. This has been accompanied by institutional action, including by the Chief Commissioner for Persons with Disabilities (CCPD), which directed health-tech platform Practo to make its website and app accessible to persons with disabilities.

In November 2024, the Supreme Court decided Rajive Raturi v. Union of India. The Court directed the government to frame mandatory accessibility rules under the Rights of Persons with Disabilities Act, 2016, distinguishing them from the broader accessibility guidelines already in place.

Following the judgment, the Department of Empowerment of Persons with Disabilities (DEPwD), together with the NALSAR Centre for Disability Studies, developed a set of accessibility rules for the Information and Communication Technology (ICT) sector in June 2025. Building on that exercise, the DEPwD has now released the rules.

Who do the Draft Rules apply to?

Nearly every technology company with Indian users. The Draft Rules cover any business that manufactures, designs, develops, or makes specified ICT items available to persons in India. This includes companies headquartered outside India. Automated systems that enable user interaction with ICT are also covered.

What products and services fall within scope?

The Draft Rules identify eight categories of ICT items. These include:

Websites, mobile and tablet apps, software, documentation and support services.

Digital content, formats and electronic documents.

ICT-based public facilities and services.

Hardware with two-way voice or video communication.

Other ICT hardware and consumer electronics for everyday use.

ICT-based consumer products.

ICT-based accessories for persons with disabilities.

ICT products used as part of public facilities and services.

There are limited exceptions, including for certain archived digital content that is no longer updated or needed for an active administrative process.

What are the key compliance obligations?

There are several. Here are the most significant ones.

Accessibility Conformance Reports (ACRs): Businesses must prepare and publicly disclose an ACR for each ICT product. The ACR must explain its compliance with the applicable Bureau of Indian Standards (BIS) accessibility standards. A senior officer, at least at Director level, must sign off on each ACR.

Platform obligations: Platforms hosting user-generated content must provide tools for users to make their content accessible, including captions, transcripts, alternative text and audio descriptions. Platforms must also guide users on creating accessible content at upload. If they don’t offer these tools, platforms have six months to build them.

Re-testing and review: Before deploying any major change affecting accessibility, businesses must re-test against the latest standards. Each ACR must also be reviewed and updated at least every two years.

Interaction with sectoral rules: Businesses must also comply with applicable sector-specific accessibility requirements. Where a sectoral regulator prescribes a stricter standard, the stricter requirement will apply.

Grievance redressal: Businesses must appoint a Grievance Redressal Officer (GRO). Complaints must be resolved within 30 days. Unresolved complaints escalate through Deputy and Chief Nodal Officers to the Chief Commissioner for Persons with Disabilities (CCPwD) or the relevant State Commissioner.

What is the compliance timeline?

Businesses with annual turnover of INR 500 crore or more get one year to comply. Those below this threshold get 18 months. Full compliance with the BIS IS 17802 standards is required within two years of notification.

What are the penalties for non-compliance?

Fines upto INR 10,000 for first contravention, INR 50,000 to INR 5 lakh for subsequent ones can be imposed under Section 89 of the RPwD Act, 2016.

Businesses must fix identified deficiencies within 90 days of a competent authority’s order. Continued non-compliance can lead to suspension or cancellation of regulatory registrations or approvals. Filing a false or misleading ACR can trigger separate action under applicable law. Sectoral regulators may impose additional requirements.

What should businesses do now?

The Draft Rules are not yet in force, but businesses can start preparing. This means mapping their digital products and services against the applicable IS 17802 accessibility standards, identifying compliance gaps, and assessing whether their platforms offer the accessibility tools contemplated by the Draft Rules.

IV. Reading reccos

Matt Levine has an accessible explanation of how Amazon’s advertising system works — and why the FTC is now suing the company over it.

Kai Williams takes a deep dive into the state of humanoid robotics, and makes the case for why humanoid robots may not catch up with humans anytime soon

Rest of World speaks with Osama Manzar about the other side of India’s data-centre boom: what rapidly expanding digital infrastructure means for the communities living alongside it.

V. What's been keeping us busy

Each month, we’ll take you behind the scenes of what’s been keeping us busy at Ikigai Law.

This month was lots of data protection. With key compliance deadlines under India’s new data-protection regime approaching in 2027, we have spent much of the past few weeks helping companies prepare. That has meant getting into the weeds of how data moves through products and businesses, assessing consent and notice requirements, reviewing contracts and internal processes, and translating regulatory requirements into changes companies can actually implement. 2027 may sound some distance away, but for businesses preparing their systems for compliance, the work has already begun.

VI. Shoutouts

Our Partner Aman Taneja is speaking at a Lexology masterclass comparing how India, China and Japan are approaching AI regulation. The session will look at the different approaches to AI governance across the three countries and what they mean for businesses navigating AI compliance in Asia. It is happening on 29 September. You can register here.

Our Partner Nehaa Chaudhari joined an ITechLaw panel on high-risk AI regulation, comparing how jurisdictions across Europe, the UK, Southeast Asia and Latin America are approaching the classification and regulation of high-risk AI systems.