News Digest — 2026-07-09

Top Stories

Hackers can use 9 of the most popular AI tools to assemble massive botnets

Ars Technica AI · 2026-07-08

Researchers have developed a new type of AI-based attack called HalluSquatting, which allows hackers to assemble large botnets by exploiting prompt injection vulnerabilities in popular AI tools. This pull-based attack enables widespread, scalable infections by luring large language models to malicious websites, unlike previous push-based methods that were limited in scope. The technique poses a significant threat to AI security and could be used for large-scale cyberattacks like DDoS operations.

Lovable reportedly in talks to double its valuation to $13.2B

TechCrunch AI · 2026-07-08

Lovable, a Swedish AI-powered vibe-coding startup, is reportedly in talks to raise $300 million at a $13.2 billion valuation, doubling its previous figure. The company, which serves founders, designers, and enterprises like Workday and Nvidia, has achieved a $500 million annualized revenue run rate and is part of a broader trend of AI-driven software startups gaining significant valuation growth.

This startup thinks robotics is about to have its ChatGPT moment

TechCrunch AI · 2026-07-08

General Intuition, a startup led by CEO Pim de Witte, proposes that robotics could follow a similar path to AI's ChatGPT moment by developing foundation models that generalize movement and interaction skills across environments, reducing reliance on massive real-world datasets. The company has built such a model using millions of hours of video game data to simulate real-world behaviors. This approach aims to shift the industry from specialized, environment-specific robot training to more efficient, broadly applicable models.

Google’s deepfake detector system used to debunk McConnell hoax pic

TechCrunch AI · 2026-07-08

Google's SynthID deepfake detection system successfully identified an AI-generated hoax image of Senator Mitch McConnell, which falsely depicted him in a hospital bed. The watermark, embedded in the image, was detected by fact-checking platforms and confirmed the image's artificial origin. The system's effectiveness highlights progress in combating AI-generated misinformation online.

Google Photos adds a new AI ‘Video Remix’ tool

TechCrunch AI · 2026-07-08

Google Photos has launched a new AI-powered 'Video Remix' feature that allows users to edit and transform videos with simple taps, using Google's Gemini Omni model. The tool offers effects like cinematic relighting, background swaps, and artistic styles such as watercolor and oil painting. It is currently available to eligible subscribers in select regions.

Technology

SpaceXAI releases Grok 4.5, which Elon describes as an ‘Opus-class model’

TechCrunch AI · 2026-07-08

SpaceXAI has launched Grok 4.5, a new large language model it describes as an 'Opus-class' model, emphasizing its ability to handle coding, research, writing, and office tasks with twice the token efficiency of leading competitors. Elon Musk highlighted its faster performance, lower cost, and competitive capabilities compared to Anthropic's Opus 4.7 in a social media post.

ChatGPT’s upgraded voice mode is better at shutting up

The Verge AI · 2026-07-08

OpenAI has introduced a new voice model, GPT-Live-1, designed to improve conversational flow by allowing real-time speaking and listening, reducing interruptions, and better handling natural dialogue. The model can switch to text-based reasoning when needed and enhances conversations with relevant visuals for topics like weather and sports.

Why this CEO thinks video games make better training data than the internet

TechCrunch AI · 2026-07-08

General Intuition, a $2.3 billion startup backed by Bezos and led by CEO Pim de Witte, argues that video game data provides superior training material for artificial general intelligence by offering rich, dynamic simulations of physical movement and spatial interactions. The company, which originated from a gaming platform, is raising $320 million with major investors including Coatue, Eric Schmidt, and DeepMind researchers, focusing on developing 'world models' that better understand real-world dynamics. Ethical concerns around potential defense applications are also being addressed in its development roadmap.