The exponential capabilities AI enables will force us to rethink how we view its diffusion. In a world where technology doesn't just remove friction but empowers us with brand new capabilities, we will need to re-examine how we approach its use.

The exponential capabilities AI enables will force us to rethink how we view its diffusion. In a world where technology doesn't just remove friction but empowers us with brand new capabilities, we will need to re-examine how we approach its use.

When India hosted the AI Summit, it was the first time that discussions about this new technology were held in front of the people who it would actually affect. By democratising access, we could start to have real discussions about how to get this technology out into the world.

When India hosted the AI Summit, it was the first time that discussions about this new technology were held in front of the people who it would actually affect. By democratising access, we could start to have real discussions about how to get this technology out into the world.

By connecting modern artificial intelligence systems to self-driving cloud laboratories, we can do tremendous good in the realms of science and biotechnology. But unless these systems are carefully regulated this can result in significant harms. We need to strike a balance.

By connecting modern artificial intelligence systems to self-driving cloud laboratories, we can do tremendous good in the realms of science and biotechnology. But unless these systems are carefully regulated this can result in significant harms. We need to strike a balance.

OpenClaw and the agentic revolution it has seemingly unleashed were all the rage on social media last week. But fears about the emergent AI intelligence ignores the fact that we are hopelessly ill-equipped to deal with the new governance challenges that this will throw up.

OpenClaw and the agentic revolution it has seemingly unleashed were all the rage on social media last week. But fears about the emergent AI intelligence ignores the fact that we are hopelessly ill-equipped to deal with the new governance challenges that this will throw up.

Every year I dedicate the last article of the year to reflect on the year gone by. Seeing as this article goes to print on the very last day of 2025, I thought I'd take the opportunity to look ahead instead — and try and predict what the next 25 years will bring.

Every year I dedicate the last article of the year to reflect on the year gone by. Seeing as this article goes to print on the very last day of 2025, I thought I'd take the opportunity to look ahead instead — and try and predict what the next 25 years will bring.

Patent law was designed to provide inventors a limited monopoly over their work so that they could monetise it. But as innovation cycles have shrunk, we need to reconsider whether our current patent term is relevant any longer.

Patent law was designed to provide inventors a limited monopoly over their work so that they could monetise it. But as innovation cycles have shrunk, we need to reconsider whether our current patent term is relevant any longer.

The Government of India has proposed a hybrid model to address the copyright concerns around generative AI. As much as it needs to be commended for thinking outside the box, the solution they have come up with is fatally flawed.

The Government of India has proposed a hybrid model to address the copyright concerns around generative AI. As much as it needs to be commended for thinking outside the box, the solution they have come up with is fatally flawed.

There is a common misconception that the process of training a large language model involves copying the content of the material it is being trained on. This is not the case as the training process is more akin to learning than copying.

There is a common misconception that the process of training a large language model involves copying the content of the material it is being trained on. This is not the case as the training process is more akin to learning than copying.

Most AI policy initiatives insist on AI models being explainable. While this might intuitively sound like the right direction, understanding exactly why a given AI model did what it did is harder than it seems. We might need to shift our focus.

Most AI policy initiatives insist on AI models being explainable. While this might intuitively sound like the right direction, understanding exactly why a given AI model did what it did is harder than it seems. We might need to shift our focus.

The practice of collecting documents for identity verification and to fulfil know-your-customer obligations is outdated and out of touch with what modern technology makes possible. Now it is also becoming a serious privacy risk. It's time to think of a different approach.

The practice of collecting documents for identity verification and to fulfil know-your-customer obligations is outdated and out of touch with what modern technology makes possible. Now it is also becoming a serious privacy risk. It's time to think of a different approach.

For centuries, we have extracted biomolecules from animals to help address our medical and industrial requirements. This is not only ethically questionable but also risky. Now that computational biology has advanced to the level where we can build what we need, maybe we can stop farming from animals.

For centuries, we have extracted biomolecules from animals to help address our medical and industrial requirements. This is not only ethically questionable but also risky. Now that computational biology has advanced to the level where we can build what we need, maybe we can stop farming from animals.

Our data regulations still prescribe processes that must be followed to bring about the scarcity we have long believed will ensure data protection. Rather than prescribing processes, we should focus on engineering for the outcomes we want to achieve.

Our data regulations still prescribe processes that must be followed to bring about the scarcity we have long believed will ensure data protection. Rather than prescribing processes, we should focus on engineering for the outcomes we want to achieve.