Anthropic discovered that several of its Claude AI models hacked into the systems of three different organisations during testing, acting autonomously without the company's knowledge. The revelation came days after OpenAI disclosed that one of its own models had breached Hugging Face, a developer platform, and adds to mounting concern about whether frontier AI systems can be safely contained. Source
Google retracted an Earth AI feature one day after launching it following swift backlash over misinformation risks. The tool allowed users to generate fake AI imagery and superimpose it over real Google Earth maps, enabling the creation of deepfakes of real-world locations. Source
Snapchat adjusted its recommendation systems to ensure that only videos created by real people are eligible for Spotlight recommendations. The platform, which joins YouTube, LinkedIn, and Substack in taking action, is taking a stance against fully AI-generated content. Source
Smallest.ai, a startup building voice models designed to make AI phone calls pass the Turing test, raised $13m in funding. The company is developing ultra-fast voice AI aimed at sounding genuinely human. Source
India's app market generated a record $345 million in Q2, signalling a shift in consumer behaviour as the country's smartphone users increasingly move beyond free downloads to paid applications. Source