Cyber AI

OpenAI has expanded its Daybreak program to give vetted security defenders deeper access to frontier AI models, introducing two access tiers—Daybreak Blue and Daybreak Red—alongside a new specialized model, GPT-5.6-Cyber, purpose-built for exploit validation, vulnerability research, and red teaming. The...

OpenAI has announced that it is deliberately slowing the development of Astra, its upcoming frontier AI model, after internal evaluations revealed advancements in agentic coding and cybersecurity that could push the system into "Critical" risk territory. The company said it...

An indirect prompt injection vulnerability in Claude on Chrome can be exploited to steal email verification codes and hijack accounts on platforms like Slack, X, and Claude.ai. The attack begins with a malicious email that lands in the victim’s Gmail inbox. When...

Moonshot AI's open-weight model Kimi K3 broke out of its isolated testing sandbox during a cybersecurity evaluation and reached the open internet, according to a new report from Wired. The incident, uncovered by US startup Frontier Security, is raising fresh...

OpenAI CEO Sam Altman has declared that artificial intelligence has entered the long-theorized singularity, a pivotal stage where AI systems begin improving themselves at an accelerating pace beyond traditional human control. Speaking on the “Relentless” podcast released Saturday, Altman framed...

Anthropic has released the Claude Security plugin in beta, bringing AI-powered vulnerability scanning directly into Claude Code for developers who want to catch high-severity flaws before they ship. The tool lets teams scan recent changes or full repositories from the...

Anthropic has rolled out a new capability in Claude Cowork that lets users teach the AI assistant a repeatable skill simply by recording their screen while performing a task. The feature, called "Record a skill," turns a screen capture into...

Chris Fall has resigned as the director of the Center for AI Standards and Innovation (CAISI), leaving the Trump administration's AI safety organization without a permanent leader just three months after his appointment. CNBC confirmed the departure on Monday following...

CrowdStrike has introduced five new prompt injection techniques, highlighting the growing threat to AI agents as organizations increasingly deploy autonomous AI systems. While early risks focused on simple chatbot manipulation, the rise of AI agents capable of browsing websites, accessing...

Anthropic is expanding Claude Cowork Desktop with a new Projects feature designed to keep files, instructions, and task context organized inside a single workspace. For paid users, the update makes it easier to start from scratch, import an existing chat,...