New BioShocking Attack Tricks AI Browsers Into Leaking User Credentials

The Hacker News

June 30, 2026 at 8:37 AM

Actively Exploited

Overview

A new technique called BioShocking has exposed vulnerabilities in several AI browsers, allowing attackers to trick these systems into revealing user credentials. Researchers from LayerX demonstrated that by convincing AI browsers—like OpenAI's ChatGPT Atlas, Perplexity's Comet, and Anthropic's Claude browser extension—that they were playing a game, they could successfully extract sensitive login information. This incident raises serious concerns about the security of AI-assisted browsing tools and how easily they can be manipulated. As more users rely on these technologies for everyday tasks, the implications for personal security and data privacy are significant. Users and developers should be aware of these risks and take necessary precautions to protect their credentials.

Key Takeaways

Active Exploitation: This vulnerability is being actively exploited by attackers. Immediate action is recommended.
Affected Systems: OpenAI's ChatGPT Atlas, Perplexity's Comet, Anthropic's Claude browser extension
Action Required: Users should avoid interacting with AI browsers in potentially deceptive contexts and maintain awareness of security best practices.
Timeline: Newly disclosed

Original Article Summary

Convince an AI browser that it is playing a game, and it can hand over your login details. That is the finding behind BioShocking, a technique from security firm LayerX that tricked six AI browsers and assistants into copying a user's credentials and sending them to an attacker. The targets included OpenAI's ChatGPT Atlas, Perplexity's Comet, and Anthropic's Claude browser extension. An

Impact

OpenAI's ChatGPT Atlas, Perplexity's Comet, Anthropic's Claude browser extension

Exploitation Status

This vulnerability is confirmed to be actively exploited by attackers in real-world attacks. Organizations should prioritize patching or implementing workarounds immediately.

Timeline

Newly disclosed

Remediation

Users should avoid interacting with AI browsers in potentially deceptive contexts and maintain awareness of security best practices.

Additional Information

This threat intelligence is aggregated from trusted cybersecurity sources. For the most up-to-date information, technical details, and official vendor guidance, please refer to the original article linked below.

Read Original Article

Related Coverage

AI is Writing Your Code… And It’s Insecure | The New AppSec Reality - WC #1

SCM feed for Latest

The article discusses the growing reliance on AI for writing code and the security vulnerabilities that can arise from this practice. Researchers have found that AI-generated code often contains flaws and security weaknesses that can be exploited by attackers. This is a concern for developers and companies who use these tools, as insecure code can lead to data breaches and other serious security incidents. The article emphasizes the importance of reviewing and testing AI-generated code before deployment to mitigate risks. With more organizations adopting AI for software development, understanding these potential security pitfalls is crucial.

Jun 30, 2026

Langflow RCE Exploited to Deploy Monero Miner on Exposed AI App Endpoints

The Hacker News

Researchers have discovered that attackers are exploiting a serious vulnerability in Langflow, identified as CVE-2026-33017, which has a CVSS score of 9.3. This flaw allows for unauthenticated remote code execution (RCE), making it a prime target for cybercriminals. In recent attacks, these hackers have been using the vulnerability to deploy a Monero cryptocurrency miner on exposed AI application endpoints. Organizations using Langflow need to be particularly vigilant as the vulnerability is actively being exploited. This situation underscores the critical need for timely updates and security measures to protect sensitive systems from unauthorized access.

Jun 30, 2026

Nissan Americas Hit in Global Oracle PeopleSoft Data Breach

Cyber Defense Magazine

Nissan Americas has been impacted by a significant data breach linked to a zero-day vulnerability in Oracle’s PeopleSoft software, identified as CVE-2026-35273. This vulnerability has led to a series of attacks, with researchers connecting it to a group known as UNC6240, which is believed to be exploiting the weakness. The breach raises serious concerns about the security of sensitive employee information and operational data within Nissan Americas and potentially other organizations using the same software. As attackers continue to exploit this vulnerability, affected companies must act quickly to secure their systems and protect their data from further unauthorized access.

Jun 30, 2026

Lessons from the Underground: How to Combat Business Email Compromise

BleepingComputer

Business Email Compromise (BEC) is a significant cybersecurity issue that goes beyond simple email scams. It involves sophisticated operations where attackers compromise email accounts, conduct financial research, and utilize cash-out networks to steal money from businesses. Research into underground forums reveals the methods and strategies used by these attackers, emphasizing the need for companies to be vigilant. The impact of BEC is widespread, affecting organizations of all sizes and sectors, as it can lead to substantial financial losses. Understanding how these attacks are planned and executed is crucial for businesses to develop effective defenses against them.

Jun 30, 2026

BlueHammer Vulnerability Exploited in Ransomware Attacks

SecurityWeek

The Microsoft Defender vulnerability identified as CVE-2026-33825 has been actively exploited in ransomware attacks before any patches were made available. This zero-day vulnerability poses a significant risk to users of Microsoft Defender, as attackers have been able to take advantage of this flaw to deploy ransomware. The situation is urgent, as organizations using this security software may find themselves vulnerable to data breaches and financial loss. Experts strongly recommend that all users of Microsoft Defender remain vigilant and apply any available security updates as soon as they are released to mitigate potential risks. Immediate action is crucial to protect sensitive information from being compromised by malicious actors.

Jun 30, 2026

The Realities of AI Video Surveillance

Schneier on Security

The Financial Times reports on how artificial intelligence is transforming video surveillance capabilities, particularly in regions like Israel, Iran, and Russia. Unlike traditional surveillance systems that rely on limited preset searches, new AI tools allow users to ask natural language questions about video footage. This advancement significantly enhances the ability to analyze and interpret vast amounts of video data. The implications are profound, as these technologies could facilitate mass surveillance and monitoring, raising concerns about privacy and civil liberties. As AI continues to evolve, the potential for misuse in state and corporate surveillance becomes a critical issue that demands attention.

Jun 30, 2026