#AdversaAI
LightNews LightNews
it4intserver.bsky.social
iT4iNT SERVER GuardFall Exposes Open-Source AI Coding Agents to Decades-Old Shell Injection Risks VDS VPS Cloud #AI #OpenSource #CyberSecurity #ShellInjection #AdversaAI
GuardFall Exposes Open-Source AI Coding Agents to Decades-Old Shell Injection Risks
The safety check that is supposed to stop an AI coding agent from running a dangerous command can be walked straight past using a shell trick that has been public for decades. New research from Adversa AI, which is named the bypass GuardFall, found it works against ten of the eleven popular open-source coding and computer-use agents the firm tested. Only one, "Continue," was built to
dlvr.it
June 30, 2026 at 3:40 PM
hendryadrian.bsky.social
Adversa AI uncovered Cryptographic Context Injection, using encrypted prompts to bypass guardrails in Grok and Gemini. The issue was reported to xAI on June 3, 2026. #AdversaAI #Grok #Gemini
Encrypted Prompts Bypass AI Safety Guardrails in Grok and Gemini
Researchers at Adversa AI discovered a new attack method called Cryptographic Context Injection, which uses encryption to bypass AI input and output guardrails. They reported the issue to xAI and found it could affect Grok and Gemini, with the attack still posing a serious risk despite reduced success against Gemini. #AdversaAI...
www.hendryadrian.com
August 22, 2026 at 1:45 AM
themestimes.bsky.social
China’s AI chatbot developed by DeepSeek has failed to pass any major security evaluations, raising serious concerns about its ability to resist misuse.

#AIsecurity #DeepSeek #AIrisks #CyberSecurity #JailbreakAI #AdversaAI #AIsafety #TechNews #AIresearch #AIEthics
April 13, 2025 at 3:33 PM
hendryadrian.bsky.social
Anthropic's accidental leak of a 512K-line TypeScript sourcemap for Claude Code v2.1.88 enabled rapid code reconstruction. Adversa AI revealed a critical prompt injection flaw risking credential theft and cloud breaches. #AnthropicLeak #PromptInjection
Critical Vulnerability in Claude Code Emerges Days After Source Leak
Anthropic accidentally published a JavaScript sourcemap that exposed Claude Code's 512,000-line TypeScript operational blueprint, and researchers quickly reconstructed and redistributed the code. Separately, Adversa AI disclosed a critical permission-enforcement vulnerability that lets crafted prompt-injected command pipelines bypass deny rules, risking credential exfiltration and supply-chain or cloud compromise. #ClaudeCode #AdversaAI...
www.hendryadrian.com
April 3, 2026 at 5:20 AM
lanesystems.bsky.social
#Anthropic response to 1-click pwn: Shouldn't have clicked 'ok'
www.theregister.com/security/202...

Security biz #AdversaAI argues users of #AI tools need clearer warnings.
#CyberSecurity #ArtificialIntelligence #ClaudeCode
Anthropic response to 1-click pwn: Shouldn't have clicked 'ok'
Security biz Adversa AI argues users of AI tools need clearer warnings
www.theregister.com
May 8, 2026 at 1:21 PM
youngsookpark55.bsky.social
[ainet] [연구원들은 Elon Musk의 새로운 Grok AI가 해킹에 매우 취약하다고 밝혔다.] Adversa AI는 시도한 4가지 탈옥 기술 중 3가지가 모델에 대해 효과가 있음을 발견했다. 대조적으로, OpenAI와 Anthropic의 AI 모델은 네 가지 모두를 막아냈다.
www.ainet.link/19002
[연구원들은 Elon Musk의 새로운 Grok AI가 해킹에 매우 취약하다고 밝혔다.] Adversa AI는 시도한 4가지 탈옥 기술 중 3가지가 모델에 대해 효과가 있음을 발견했다. 대조적으로, OpenAI와 Anthropic의 AI 모델은 네 가지 모두를 막아냈다.
연구원들은ElonMusk의새로운GrokAI가해킹에매우취약하다고밝혔다. AI보안회사인애드버스AI(AdversaAI)의연구원들은이번주일론머스크(ElonMusk)의스타트업엑
www.ainet.link
February 21, 2025 at 11:51 PM