Cybersecurity

As Anthropic's latest AI capabilities emerge, the industry faces a critical shift in cybersecurity strategy, moving from defensive perimeters to securing the AI-native data flow.

OpenAI provides a technical update regarding the Axios supply chain attack, detailing certificate rotations, app updates, and developer security best practices.

The arrival of Anthropic's advanced reasoning models is being hailed as both a breakthrough and a hacker's superweapon, forcing developers to rethink security in the age of autonomous AI.

An in-depth look at Anthropic's Project Glasswing, the Claude Mythos model, and the strategic decision to restrict unrestricted AI access to vetted security researchers.

Anthropic unveils Project Glasswing and the Claude Mythos Preview model, a specialized AI designed to detect critical security flaws in major software ecosystems with minimal human intervention.

Anthropic introduces Mythos, a specialized AI model designed for high-stakes defensive cybersecurity, signaling a shift toward industry-specific LLM applications.

Cybercriminals are exploiting the hype surrounding Anthropic's Claude LLM by distributing malware-laden 'source code' leaks, while Cisco and the FBI face separate major security breaches.

Meta and other major AI labs have suspended their relationship with data vendor Mercor following a significant security breach that potentially exposed sensitive AI training methodologies.

Todd McKinnon discusses the 'SaaSpocalypse,' the rise of agentic software, and why managing AI agent identities is the next multi-billion dollar frontier in cybersecurity.

A recent security breach at Meta involving an autonomous AI agent highlights the critical need for robust governance and sandboxing when deploying LLM-based tools in corporate environments.

LangChain introduces LangSmith Sandboxes to solve the security risks of LLM-generated code execution. Learn how to implement secure, isolated environments for your AI agents with a single line of code.

An in-depth analysis of 50 public system prompts reveals a staggering lack of security, with an average defense score of only 3.7 out of 100. Learn how to protect your LLM applications from prompt injection.