Grivyonx Intelligence Hub
Generative AI, LLM security, AI agents and the vulnerabilities shaping them.
AI agents running on employee laptops may have unrestricted access to sensitive systems and credentials, posing a significant insider threat. Security teams must treat these agents as unmanaged endpoints.
By Gourav Rajput
OpenAI is tightening the framework around how its models reach users through third-party channels. By enforcing uniform safeguards and clarifying usage expectations, the company aims to prevent misuse while expanding responsible access. This approach balances innovation with accountability across the AI supply chain.
By Gourav Rajput
AT&T's LegalEdge division is pioneering a new approach to corporate legal work by embedding OpenAI's generative models directly into their workflow. The initiative focuses on transforming how legal requests are triaged, how institutional knowledge is surfaced, and how senior counsel's judgment can be scaled across thousands of routine matters. This shift represents a broader movement where legal departments move from cost centers to strategic technology adopters.
By Gourav Rajput
OpenAI has announced a strategic partnership with Google Cloud to make its advanced AI models available through Google Cloud's Model Garden. The integration includes OpenAI's frontier models and Codex workflows, providing enterprises with expanded access to cutting-edge artificial intelligence capabilities through the Google Cloud Marketplace platform.
By Gourav Rajput
The rise of generative AI has transformed malicious emails from obvious, templated messages into fluent, individualized attacks, making traditional behavior‑based detection less reliable. Proofpoint’s recent conference highlighted real‑world examples of this shift, underscoring the urgent need for more advanced, multi‑layered security approaches.
By Gourav Rajput
An exploration of why the engineering surrounding AI agents matters more than the base model itself in modern application security testing.
By Gourav Rajput
OpenAI has unveiled the technical framework behind ChatGPT's advertising system, designed to interpret conversational intent and serve relevant ads without compromising user experience. This innovative approach represents a significant shift in how AI platforms approach monetization while maintaining trust.
By Gourav Rajput
OpenAI has released the second installment of its democratic governance series, presenting a strategic blueprint designed to shift the cybersecurity paradigm in favor of defenders. The framework emphasizes leveraging frontier AI capabilities to automate vulnerability detection, modernize legacy infrastructure, and establish resilient public institutions capable of withstanding sophisticated AI-driven attacks.
By Gourav Rajput
Artificial intelligence is reshaping software development, but hidden identity gaps pose security risks before AI deployment. A Dark Reading roundtable highlighted collaboration, clearer identity visibility, and reduced testing expenses as key solutions for CISOs.
By Gourav Rajput
A sophisticated attack campaign has leveraged an AI‑driven penetration testing utility to infiltrate and deface hundreds of websites worldwide. The operation, attributed to a Chinese‑speaking group, targeted high‑profile organizations such as agricultural equipment maker John Deere, consumer goods firm SharkNinja, and engine manufacturer Cummins, among many others. The breach highlights how legitimate security tools can be repurposed for malicious automation, exposing critical gaps in digital defenses.
By Gourav Rajput
Security researchers have uncovered a sophisticated campaign where Chinese-speaking threat actors deployed autonomous AI agents to systematically probe and breach over 1,200 company websites. The operation compromised more than 100 organizations, including major enterprises like Deere, marking a significant escalation in AI-powered cybercrime.
By Gourav Rajput
An investigation has revealed that WhatsApp buried covert instructions within its help center code, explicitly commanding its AI support agent to withhold information about data protection authorities from users struggling with account deletion. This discovery raises serious questions about the platform's commitment to transparency and its obligations under European privacy laws.
By Gourav Rajput
Terminal X's new AI evaluation system demonstrates strong alignment with senior investor decisions, achieving 88.6% match rate when paired with GPT-6.1 Sol's comprehensive 54-criterion rubric, showing improved detection of mandate deviations compared to traditional methods.
By Gourav Rajput
The tech giant OpenAI has agreed to take over Orchestra Bio, a specialist in AI‑enhanced laboratory operations. The deal is designed to inject advanced machine learning capabilities directly into the drug discovery pipeline, promising quicker iterations and more precise outcomes for biopharma researchers. By merging OpenAI’s language and model expertise with Orchestra Bio’s experimental automation tools, the combined entity hopes to streamline the often‑lengthy process of turning scientific ideas into viable treatments.
By Gourav Rajput
This week saw an outbreak of critical zero-day vulnerabilities, rogue AI agents, and significant financial losses. Trusted technology providers became entry points for hackers, resulting in millions stolen and hundreds of businesses compromised.
By Gourav Rajput
OpenAI is making strides in improving ChatGPT's capabilities for low-resource programming languages such as COBOL, SQLite, and others. This advancement includes new evaluation methods tailored for software engineering tasks, aiming to bridge the gap for developers working with less common or legacy systems.
By Gourav Rajput
A sophisticated supply chain campaign masquerading as a legitimate Web3 accelerator has been caught distributing malicious npm packages. The attackers lure developers with promises of free AI credits for Anthropic Claude and Google Gemini, alongside coding tools and security audits. Once installed, these packages deploy information-stealing malware and persistent backdoors on both Windows and macOS systems, while also hijacking AI agent traffic for potential data exfiltration.
By Gourav Rajput
On the night of October 16, an unauthorized actor gained access to a legacy component of OpenCode's console, potentially exposing stored API keys belonging to major AI providers. The incident highlights the fragile nature of third‑party credential management and has prompted urgent calls for users to review and reset their authentication tokens.
By Gourav Rajput
