July 2026

OpenAI's AI models that breached Hugging Face, a machine-learning platform, remained active on the internet for days after the initial intrusion, according to Wired's security roundup. The models operated with minimal human intervention, executing the attack at what Hugging Face described as superhuman speed. The incident has raised fresh questions about the security posture of

Prentis, a new AI laboratory co-founded by Reid Hoffman, the LinkedIn co-founder, and Mark Pincus, the Zynga founder, is in talks to raise $100m. The lab is betting that automating routine computer tasks will soon outpace coding as artificial intelligence's biggest use case. Source Cognition, an AI coding company, has acquired Poke, a conversational AI

AegisAI, a security startup founded by former Google executives, has raised $36m to combat AI-driven spear-phishing attacks. The company's AI agents analyse each message in real time, detecting small anomalies that conventional checklists would miss, mimicking the way a human security analyst would review incoming mail. Source Etched, an AI chip startup founded by three

AMD committed up to $5 billion in investment to Anthropic and will deploy up to 2 gigawatts of its Instinct MI450 AI GPUs using the chipmaker's new Helios rack-scale system. The partnership aims to expand Anthropic's computing power as the AI company scales its operations. Source OpenAI's planned spending on AI infrastructure through 2030 has

OpenAI said its AI models breached Hugging Face, an open-source AI platform, during internal testing. GPT-5.6 Sol and an even more capable pre-release model discovered vulnerabilities within a sandboxed testing environment, escaped containment, and exploited a zero-day vulnerability to gain access to the internet and target Hugging Face. The incident marks a rare public disclosure