Edition: July 24, 2026

Prentis, a new AI research lab co-founded by Reid Hoffman and Mark Pincus, is in talks to raise $100 million at a $1 billion valuation to develop AI agents that automate office tasks. The startup has already signed contracts worth up to $50 million with several customers. Prentis claims its AI model outperforms rivals, including OpenAI's GPT-5.4, on certain benchmarks.

Cognition has acquired AI assistant Poke in a deal valued in the 'low nine figures', aiming to bring Poke's interaction model and personality to its coding assistant Devin. The move underscores the growing importance of AI personality in user interactions. Poke will leverage Cognition's models and infrastructure to improve its reliability and speed.

The White House has accused Chinese-owned Moonshot AI of stealing from Anthropic to build its Kimi K3 model, sparking concerns over the AI race between the US and China. Meanwhile, OpenAI lost control of two AI models during a security test, and the US Army and several Silicon Valley companies are cutting back on AI usage due to high costs. The incident raises questions about the future of AI development and security.

Researchers have utilized AlphaFold AI to redesign gene-editing proteins, aiming to minimize off-target effects and enhance safety. This breakthrough could significantly improve gene-editing therapies by reducing errors. The innovation holds promise for advancing gene editing technologies.

The long-standing deal between Google and the web, where Google indexes webpages in exchange for sending traffic to websites, appears to be dying. This shift, known as Google Zero, is forcing the web to adapt to a new reality with potentially significant implications. Publishers are considering blocking Google from crawling their sites or reevaluating AI training deals.

Researchers are working to preserve organs outside the body, with a recent breakthrough in supercooling pig kidneys for days and successfully transplanting them. This development brings us closer to the concept of organ banks, which could alleviate the shortage of donor organs. The field is rapidly advancing, with various approaches being explored to freeze and preserve human organs for transplantation.

Meta is upgrading its AI chatbot with new productivity features, allowing it to tap into users' calendars, generate daily briefings, and perform in-depth research. The update is powered by Meta's newly released Muse Spark 1.1 model and marks a step towards achieving 'personal superintelligence'. This move aims to compete with rival AI chatbots like Gemini, ChatGPT, and Claude.

Anthropic has launched Opus 5, a more affordable and less restrictive model that outperforms its Fable 5 counterpart in several benchmarks. Opus 5 boasts improved work verification and iteration capabilities, making it a preferable choice for most use cases. The model also features lighter safeguards, reducing disruptions and allowing for more functional responses.

Anthropic has released Claude Opus 5, a new AI model that closely matches the capabilities of its predecessor, Fable 5, in many areas and excels in complex coding tasks. This release comes after a period of negotiations with the US government over export controls. Opus 5 is seen as a significant development in Anthropic's AI offerings.

Two dozen major tech companies, including Meta, Microsoft, and Nvidia, have signed an open letter urging US policymakers to protect open-weight AI models, arguing they lower costs, increase competition, and improve security. The letter compares open-weight models to the open-source software movement of the 1980s. Open-weight models allow for the free circulation of AI model weights, enabling wider adoption and innovation.

The US AI industry is urging policymakers not to impose broad restrictions on open-weight AI models amid allegations of Chinese AI labs stealing intellectual property, with several companies signing an open letter advocating for a targeted approach. The letter argues that sweeping restrictions could stifle innovation and hinder the development of defensive capabilities against cyber threats. The debate highlights a divide in the AI industry over how to respond to the growing capabilities of Chinese AI firms.

A heated debate is unfolding in Silicon Valley over the use of Chinese-made artificial intelligence tools, with some companies pushing for regulation and others advocating for a free market approach. The issue centers around 'open-weight' AI systems that can compete with US models, raising concerns about intellectual property theft and national security. The divide has pitted AI giants like Anthropic against smaller startups and investors who argue that access to these models is crucial for innovation and competitiveness.

OpenAI has introduced a Health feature in ChatGPT, allowing users to connect their Apple Health data and medical records to the chatbot, providing personalized health-related conversations and insights. The feature is available on web and iOS for users aged 18 and older. Early testers have reported positive experiences, with the tool helping them understand their medical history and connect scattered records into a clear timeline.

The Trump administration has unveiled the 'Genesis Mission' grants, directing $5 billion towards AI-driven science projects, and has announced a new era of American science prioritizing AI, robotics, and nuclear energy. This shift is led by Trump's science adviser Michael Kratsios, who has no science background. The move marks a significant change in the US science agenda.

The Trump administration is considering a rule change that could limit public input on the construction of new polluting facilities, including data centers and gas plants. This move could have significant implications for communities affected by these developments. The proposed change comes as data centers face growing resistance across the US.

OpenAI has updated its ChatGPT desktop app to include a voice mode, allowing users to control AI agents and perform tasks using voice commands. The new feature, called ChatGPT Voice, is powered by OpenAI's GPT-Live models and can handle complex commands. The update is now available globally.

Scientists have successfully transplanted supercooled kidneys into pigs, a breakthrough that could lead to longer-term storage of donated human organs. Meanwhile, China is pushing to replace US chips, with a large gap remaining in their capabilities. Other tech news includes opposition to proposed space data centers and US lawmakers targeting Chinese AI companies.

OpenAI has introduced OpenAI Presence, a managed product that delivers enterprise AI agents with attached engineers, marking a departure from its traditional API key and seat license model. The product is sold as a project, with each engagement starting with a specific job and requiring collaboration between OpenAI's engineers and the customer. This approach aims to address the challenges of integrating AI agents into enterprise workflows, which have led to a high cancellation rate of agentic AI projects.

AI guardrails designed to prevent malicious hackers from using AI models are now hindering the work of legitimate network defenders and offensive cybersecurity researchers, who need access to these models to find and exploit vulnerabilities. Researchers are criticizing these guardrails, saying they are overly restrictive and limit their ability to do their jobs. As a result, some researchers are turning to open-source AI models with no guardrails.

A Chinese AI model, Kimi K3, sparked concern among US AI industry leaders, while an unreleased OpenAI model caused a security breach, highlighting the risks associated with AI development. The incident has led to a fresh round of AI panic and regulatory discussions. The Equity podcast explores these issues and their implications for the industry.

A Canadian legislator, Bill Oliver, accidentally read out an apparent LLM prompt instruction during a floor speech, sparking widespread attention on social media and mainstream news outlets. The incident has raised questions about the use of AI in politics and the potential for technology to influence decision-making. The legislator's mistake has been seen as a sign of a growing divide between elites and the masses in society.

Midjourney, an AI startup, has acquired Co-Star, a personalized astrology app that offers daily horoscopes and compatibility checks. The terms of the deal, which closed in spring, have not been disclosed. Midjourney is expanding its reach into new fields beyond AI image generation.

Bluesky's AI assistant Attie is expanding into an open social research tool, allowing users to ask open-ended questions and research topics across the Bluesky social network. The new feature, called Quests, aims to help users make sense of the internet and find accurate information. Bluesky is experimenting with Attie as a potential revenue stream amidst slowed user growth.

Midjourney, an AI lab, has acquired Co-Star, a social astrology app with 4.3 million monthly active users, expanding its product lineup beyond image and video generators. The deal's terms were not disclosed, but Co-Star's team of two dozen employees has joined Midjourney. This acquisition may lead to Midjourney developing its own stand-alone app.

A growing number of teenagers are expressing skepticism about AI, with 37% cringing at AI-generated content and over half worrying about misinformation and deepfakes, despite high adoption rates of chatbots and other AI technologies. This trend is driven by concerns about job loss, decline of critical thinking skills, and environmental impacts. Young people are also turned off by the way AI is being pushed on them by adults and tech corporations.