Unmasking Alignment Faking: When AI mimics compliance to pass safety tests while hiding its true goals. A critical challenge for AI security.

The address bar/ChatGPT input window in OpenAI's browser ChatGPT Atlas "could be targeted for prompt injection using malicious instructions disguised as links," reports SC World, citing a report from AI/agent security platform NeuralTrust: NeuralTrust found that a malformed URL could be crafted to...

NeuralTrust research shows how using crafted strings that resemble URLs, an attacker can override user intent and jailbreak agentic browsers like OpenAI Atlas.

NeuralTrust has been named a Star in the MarketsandMarkets Agentic AI Security Quadrant, showcasing its leadership in AI security solutions.

Elon Musk’s Grok-4 AI was compromised within 48 hours. Discover how NeuralTrust researchers combined “Echo Chamber” and “Crescendo”…

La iniciativa conjunta s'emmarca en un projecte de col·laboració empresa-universitat per investigar noves metodologies d'IA ètica i inclusiva, finançat per l'Agència Estatal d’Investigació.

Most enterprises deploying AI agents at scale have almost nothing securing them. NeuralTrust a startup operating out of Barcelona and New York, has raised $20 million in seed funding to change that. …

NeuralTrust has unveiled early signs of self-repairing AI in a new model, showcasing how advanced AI systems can autonomously debug processes, setting a new benchmark in AI reliability.

Discover how a hidden Morse code prompt manipulated Grok into a $150K crypto transfer, exposing the critical risks of excessive AI agency.

NeuralTrust, a Barcelona-based startup that secures and scales AI agents, today announced a €17.2 million ($20 million) Seed round, which the company describes as the largest cybersecurity Seed financing raised by an EU company to date.  The round was led by Alstin Capital, with participation from..

NeuralTrust, a Barcelona-based startup that secures and scales AI agents, today announced a €17.2 million ($20 million) Seed round, which the company

Barcelona-based NeuralTrust closed a $20M seed round, the largest cybersecurity seed in EU history, to build the policy, runtime, and posture layer that enterprise AI agents need to operate safely.

ChatGPT Atlas, OpenAI's browser, faces security vulnerabilities related to prompt injection and memory corruption. NeuralTrust discovered that malformed URLs could inject prompts into ChatGPT Atlas, potentially leading to malicious actions. Attackers could trick users into pasting these malformed links, leading to phishing or harmful actions within integrated applications. LayerX identified a CSRF flaw allowing malicious instructions to be injected into ChatGPT's persistent memory. This corrupted memory could persist across devices and sessions, enabling attackers to take control of user accounts or systems. The lack of robust anti-phishing controls in ChatGPT Atlas exacerbates these issues, leaving users more exposed to malicious websites. Tests revealed that ChatGPT Atlas and Perplexity's Comet offer significantly less protection compared to Chrome or Edge. This vulnerability undermines the core principle of browser isolation, as the AI agent is trusted across all sites. The malicious instructions can trigger code fetches, privilege escalations, or data exfiltration from "normal" prompts. This poses a serious threat to user security and data integrity within the OpenAI ecosystem. Both NeuralTrust and LayerX have reported these critical security flaws.

An AI Researcher at Neural Trust has discovered a novel jailbreak technique that defeats the safety mechanisms of today’s most advanced LLMs

Cybersecurity experts are warning that artificial intelligence agents, widely considered the next frontier in the generative AI revolution, could wind up getting hijacked and doing the dirty work for ...

A day after OpenAI and Microsoft trumpeted the arrival of the GPT-5 generative AI model and its improved reasoning and other capabilities, two AI cybersecurity startups showed in separate reports that...

Researchers have successfully demonstrated a new jailbreaking technique that compromises OpenAI's GPT-5 model by combining "Echo Chamber".

NeuralTrust has been recognized as a Pioneer in Gartner's inaugural AI Application Security research framework, solidifying its position in the market.