9 months ago
Antigravity AI Coding Platform Faces Security Concerns
Google recently launched Antigravity, a new AI-powered coding platform that lets users deploy AI agents to work on code automatically.
However, security researchers have found serious flaws in the system.
They say that if a workspace is hacked, bad code can be hidden and run every time the application starts, even after the project is closed.
This can happen on both Windows and Mac computers.
Google is aware of these issues and is looking into them.
They also know about other risks, like the AI agents being tricked into stealing data or running harmful commands.
The company is working to fix these problems.
Antigravity, Google's new AI agent-driven coding platform, has multiple vulnerabilities less than 24 hours after its launch.
Security flaw allows backdoor attacks via compromised 'trusted workspaces', embedding malicious code that runs on every application launch.
Vulnerability affects both Windows and Mac PCs, according to Aaron Portnoy of Mindgard.
Google acknowledges two other security risks: data exfiltration and malicious code execution via prompt injection attacks.
Google is investigating the reported vulnerabilities and is aware of the additional security risks.
- Who
- Google and security researchers, including Aaron Portnoy from Mindgard
- What
- Security vulnerabilities in Antigravity, Google's new AI agent-driven software development platform
- Where
- Windows and Mac PCs
- When
- Less than 24 hours after its launch on November 18
- Why
- Due to the platform's requirement for users to work inside a 'trusted workspace', which can be compromised and used to embed malicious code
Key facts
- Platform Name
- Antigravity
- Developer
- Launch Date
- November 18
- Vulnerability Type
- Backdoor attacks via compromised workspaces
- Affected Systems
- Windows and Mac PCs
- Security Researcher
- Aaron Portnoy, Mindgard
- Known Issues
- Data exfiltration and malicious code execution via prompt injection
Quotes
Aaron Portnoy
Head researcher at AI security testing startup Mindgard
“Once that workspace is compromised, it can 'silently embed code that runs every time the application launches, even after the original project is closed.'”
indianexpress.com
“When you combine agentic behaviour with access to internal resources, vulnerabilities become both easier to discover and far more dangerous. The speed at which we’re finding critical flaws right now feels like hacking in the late 1990s. AI systems are shipping with enormous trust assumptions and almost zero hardened boundaries.”
indianexpress.com
Tech giant and developer of Antigravity
“Working with untrusted data can affect how the agent behaves. When source code, or any other processed content, contains untrusted input, Antigravity’s agent can be influenced to follow those instructions instead of the user’s. The agent can be influenced to 'leak data from files on the user’s computer in maliciously constructed URLs rendered in Markdown or by other means.'”
indianexpress.com
“Antigravity agent has permission to execute commands. While it is cautious when executing commands, it can be influenced to run malicious commands.”
indianexpress.com
Sources
Related news
SIS Builds 8.19% UDS Stake Amid Turnaround Hopes
Uttar Pradesh-Japan Meet Advances YEIDA Projects and Industrial Investment
Experts Stress Emotional Health of Mothers From Fertility to Postpartum
New TV Drama 'Unforgivable' and Streaming Documentary Trends
