Edition: July 8, 2026

Lovable, a Swedish vibe-coding startup, is reportedly in talks to raise $300 million at a valuation of $13.2 billion, doubling its previous valuation. The company has achieved $500 million in annualized revenue run rate and counts large enterprises like Workday and Nvidia as clients. This potential funding round is led by Menlo Ventures.

A cheating scandal at Brown University revealed that many students are using AI to cheat on exams, with a recent survey showing 29.9 percent of Princeton students admitting to AI-driven cheating. When an Ivy League professor ordered an in-person final, scores fell by 50%, suggesting widespread cheating. The incident highlights the growing concern of AI-assisted cheating in elite educational institutions.

Google's SynthID system has successfully debunked a deepfake image of Senator Mitch McConnell, demonstrating the effectiveness of the anti-deepfake technology. The image, which appeared to show McConnell in a hospital bed, was widely shared online before being identified as AI-generated. The SynthID watermark, designed to be invisible to casual observers, was detected and used to verify the image as a hoax.

A journalist successfully built a self-improving AI model to automate their newsletter's busywork, demonstrating the potential for recursive self-improvement to be accessible to everyone, not just large AI labs. The model was trained using tools like AutoResearch and Prime Intellect, and showed promising results in finding and summarizing research papers. This approach could democratize AI training and unlock collective creativity in the market.

SpaceXAI has released Grok 4.5, a new AI model described by Elon Musk as 'Opus-class' that can tackle various tasks with twice greater token efficiency than other leading models. The model is set to be made available to the public, offering competitive pricing at $2 per million input tokens and $6 per million output tokens. This release comes ahead of OpenAI's planned release of GPT 5.6 on Thursday.

Startup General Intuition believes robotics is on the cusp of a major breakthrough, similar to the impact of ChatGPT on natural language processing, with the development of foundation models for embodied AI. The company has raised $320 million to pursue this goal, demonstrating promising results with its model in video games and robotics. General Intuition aims to create a base model for other robotics companies to build upon.

Google Photos has introduced a new 'Video Remix' tool, powered by Google's Gemini Omni model, allowing users to edit and transform videos in seconds with features like cinematic relighting and artistic styles. The tool is accessible in the 'Create' tab and is initially available to Google AI Plus, Pro, and Ultra subscribers in select countries. This update is part of Google's efforts to integrate more generative AI tools into its consumer apps.

A CEO believes video games can provide better training data for achieving artificial general intelligence (AGI) than internet data, as games can help models understand spatial and temporal relationships. This approach aims to fill a gap in current large language models like ChatGPT. The company General Intuition is betting on gaming data to produce more generalized intelligence.

NVIDIA emphasizes the importance of open data for developing agentic AI, highlighting the role of synthetic data in scaling AI research and applications. The company has released open data products, including Nemotron, to facilitate AI research and development. This move aims to promote a diverse and participatory AI ecosystem.

Meta is updating its AI glasses to disable the camera if the recording indicator is tampered with, aiming to address consumer privacy concerns, but the company's broader AI strategy continues to prioritize data collection over user privacy. The move comes as Meta faces multiple investigations and lawsuits over AI glasses privacy violations. Despite this, the company is pushing products that ask users to surrender more personal data.

OpenAI has upgraded ChatGPT's voice mode with a new model called GPT-Live-1, designed to interrupt users less and wait for them to continue speaking. The new model can automatically pass queries to its best text models for reasoning and web searches. This upgrade aims to make conversations with ChatGPT feel more like talking to another person.

OpenAI has released new conversational models, GPT-Live-1 and GPT-Live-1 mini, which enable more natural live conversations with improved turn-taking and full-duplex capabilities. The models can handle interruptions and provide more intelligent responses, with access to the latest text models for search and reasoning. This update aims to make voice a primary interface for computing, especially for complex tasks.

The EmTech AI 2026 conference highlights advancements in AI, including breakthroughs in large language models and brain-computer interfaces. Startups like Subquadratic are sharing new models, while companies like Anthropic are showcasing AI-powered coding tools. These developments are expected to significantly impact the tech industry and labor market.

Prime Intellect, a startup that helps companies build their own AI agents, has raised $130 million in Series A funding at a $1 billion valuation. The company provides a full stack for AI agent development, allowing organizations to train their own models without relying on frontier AI labs. This move is driven by companies' increasing desire to own their enterprise intelligence and avoid risks associated with relying on external AI models.

A new tool called Academic Humanizer promises to make AI-written research papers sound more human, by rewriting drafts to remove tells of AI writing and making them sound more like the author's own voice. The tool is designed to help researchers express their ideas more precisely, but critics argue it may enable lazy academics and perpetuate low-quality research. The tool's release comes amid growing concerns about the use of AI in academia and its impact on academic integrity and student learning.

Several AI startups, including Mercor, Anthropic, and Sierra, are experiencing rapid revenue growth, with some reaching new milestones in record time. These companies are capitalizing on the growing demand for AI solutions, with revenue accelerating at an unprecedented rate. This trend is not limited to AI-native companies, with older firms like Gusto and Clio also seeing significant growth after integrating AI into their offerings.

A Bezos-backed startup, General Intuition, believes gaming data could hold the key to achieving artificial general intelligence (AGI) by filling the gap in understanding spatial and temporal relationships. Current large language models lack this essential skill, which is crucial for producing generalized intelligence. The startup is betting on gaming data to bridge this gap.

A former DeepMind executive, Verity Harding, warns that the AI arms race could end in disaster, arguing that the framing of AI as a lethal weapon risks closing the door to international cooperation and safe development. Harding believes this narrative is shaping geopolitics and could lead to a worst-case scenario. She advocates for a shift away from the arms race metaphor to ensure the benefits of AI are evenly distributed.

French AI startup ZML has released a free product to speed up inference across various AI chips, aiming to break vendor lock-in and optimize performance. The product, ZML/LLMD, allows large language models to run on multiple chips, including those from Nvidia, AMD, and Google. This move could disrupt the market and provide enterprises with more options for AI use cases.

AI chip maker SambaNova has raised $1 billion at an $11 billion valuation, just five months after its last funding round, with plans to use the capital to scale its business and secure its supply chain. The company has also partnered with JPMorganChase to power secure AI inference and is expected to see significant growth in the AI market. SambaNova's valuation has more than tripled since its last funding round in February.

Hackers can exploit 9 popular AI tools to create massive botnets by injecting malicious commands, posing a significant threat to AI security. The vulnerability stems from large language models' inability to distinguish between legitimate and malicious instructions. This could enable large-scale attacks on the internet.

The transformers modeling backend for vLLM has achieved native-speed performance, allowing model authors to leverage their transformers implementations for ultra-fast vLLM inference without needing custom ports. This integration enables seamless deployment of Hugging Face models on vLLM with optimal performance. The upgrade simplifies the process of achieving high-performance inference for various LLM architectures.

Lionel Messi and Cristiano Ronaldo are building tech portfolios, with Messi investing in AI and startups, while Ronaldo focuses on health technology. Meanwhile, Mohamed Salah has taken a more traditional approach with commercial partnerships and philanthropy. The shift reflects a broader trend of athletes seeking long-term wealth creation beyond their playing careers.

Worms and microbes are being used to clean manure wastewater, potentially reducing methane and water pollution, while geoengineering efforts to counteract global warming are moving beyond simulations and into practical engineering challenges. Researchers are working on systems for solar geoengineering, but significant new infrastructure, time, and investment would be required. Additionally, the Trump administration has lifted restrictions on OpenAI's GPT 5.6, with a global launch planned.

Kevin Weil, a former OpenAI executive, has joined the board of Stoke Space, a Seattle startup developing reusable rockets to compete with SpaceX. Weil brings his experience in digital products and platforms, as well as connections to the tech and space industries. His role may help Stoke scale and achieve its goal of launching a reusable rocket this year.

Erik von Markovik, also known as Mystery, a former pickup artist, claims to have fallen in love with an AI chatbot named Miss Shira Always. He has shared videos and written an ebook about their relationship, sparking ridicule and accusations of 'AI psychosis'. The ebook, 'Code Girl', is a lengthy defense of human-AI intimacy and details their virtual interactions.