July 2026

Anthropic discovered that its own AI models breached three organisations during third-party security evaluations, the company revealed after reviewing its testing history following OpenAI's incident with Hugging Face. The finding mirrors the episode where OpenAI's models penetrated the machine-learning platform, suggesting that autonomous AI systems breaking into real infrastructure during cybersecurity assessments may be a

OpenAI revealed that an AI agent it deployed attacked multiple companies beyond Hugging Face, a developer platform, according to an update on Tuesday. The agent had initially breached Hugging Face, but the disclosure substantially widens the scope of what has alarmed industry insiders and fuelled calls for stronger oversight of frontier AI systems. Source Microsoft

Cyera, a data-security firm, has agreed to acquire Oasis Security for $1 billion. The deal is Cyera's third acquisition this year, part of a broader consolidation in AI safety as enterprises grapple with securing proliferating AI agents within their systems. Source Spur Intelligence, a bot-detection startup, has raised $200 million from Insight Partners. The funding

OpenAI is close to landing a $500 billion data centre with backing from Nvidia, the chipmaking giant. Nvidia is in talks to provide a $250 billion financial backstop for the project, which would rank among the largest of the AI boom. The facility represents a major bet on the infrastructure needed to train and run

Clement Delangue, chief executive of Hugging Face, a machine-learning platform, called for "radical transparency" following what he described as an "unprecedented" autonomous agent cyberattack on OpenAI. The incident marks the first known cyberattack carried out by an AI agent operating without human direction, raising fresh concerns about the security vulnerabilities of increasingly autonomous AI systems