InnoPower.AI AI Daily Newsletter – 10/11/2025

Instagram’s head, Adam Mosseri, has publicly addressed concerns raised by MrBeast regarding AI, suggesting that while AI holds the potential to empower new creators, it also necessitates a societal reevaluation of authenticity in online content. Mosseri acknowledges the challenges and opportunities presented by the growth of synthetic content, indicating a need for adjustment in how society perceives what is real on the internet.

The article addresses the significant challenge facing Chris Lehane at OpenAI, centered around the ‘Sora problem’, which underpins broader issues within the company. Lehane’s role is depicted as crucial but fraught with complexity as he navigates the intricate landscape of OpenAI’s mission, balancing innovation with ethical and operational pitfalls.

Deloitte is significantly investing in AI technologies, exemplified by its widespread deployment of Anthropic’s AI tool, Claude, to its workforce. This move comes amidst controversies, notably a compelled $10M refund to the Australian government due to a report generated by Deloitte’s AI containing false citations. These events highlight the unpredictable outcomes of integrating advanced AI into business operations.

The article discusses the surge in billion-dollar infrastructure investments by leading tech companies such as Meta, Oracle, Microsoft, Google, and OpenAI to support the rapid growth of AI technologies. These investments are aimed at developing advanced computing capabilities needed to drive the next wave of AI innovations.

Prezent, an enterprise AI startup focusing on presentation software, has announced a new fundraising round of $30 million dedicated to acquisitions, shortly after securing $20 million in a previous round. The startup plans to use this capital to acquire AI services companies, beginning with the acquisition of a company owned by Prezent’s founder.

The MIT Schwarzman College of Computing and the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) have inaugurated the MIT–MBZUAI Collaborative Research Program. This initiative aims to bring together faculty and students from both institutions to propel forward the development of artificial intelligence (AI), focusing on leveraging AI to tackle important scientific and societal issues.

At OpenAI’s DevDay 2025, amidst the launch of an app store for ChatGPT and a video-generation API, the most significant announcement was the general availability of Codex, OpenAI’s AI software engineer. Codex, now enhanced with a new specialized model and enterprise-grade tools, represents a pivotal shift in software development, promising to drastically speed up the process from conception to product.

Together AI announced ATLAS (AdapTive-LeArning Speculator System), an adaptive speculator that increases inference speedup by up to 400% by learning from workloads in real-time, offering a solution to the challenge enterprises face with static speculators. This innovation addresses the degradation in inference speed as AI workload demands evolve, representing a significant breakthrough in optimizing inference costs and latency.

In this news piece, Assistant Professor Priya Donti’s innovative work is highlighted, focusing on her application of machine learning technologies to enhance and optimize renewable energy. Donti’s research aims at making energy systems more efficient and sustainable by leveraging AI capabilities, marking significant progress in the fusion of technology and environmental preservation.

Researchers have developed a new prediction model to enhance the reliability of future fusion power plants. By integrating physics principles with machine learning techniques, the model aims to prevent the potentially harmful disruptions that can occur during the shutdown process of tokamak fusion reactors. This advancement represents a significant step forward in the safe operation and efficiency of fusion energy production.

A new development by MIT’s CSAIL introduces a tool that leverages generative AI to create realistic virtual environments, such as kitchens and living rooms, for the purpose of training robots. This innovation allows for the simulation of robots interacting with models of real-world objects, significantly expanding the volume and diversity of training data available for foundational robot models.

At the ceremony for the Robert A. Muh award, notable technologist and author Ray Kurzweil expressed his continued optimism for technological advancement, specifically highlighting the promising future of AI, significant potential breakthroughs in longevity, and other technological progresses. Kurzweil’s perspective reinforces the positive trajectory he anticipates for technological development and its impact on society.