Monitor this situation.

Unsubscribe anytime.

[SITUATION] · [QUIET] · [TECHNOLOGY]

2 clusters · 9 sources · 1 days · First seen 22 days ago · Last updated 22 days ago

xAI Grok chatbot technical and security issues

Overview

xAI’s Grok chatbot has faced technical and security challenges. Initially, the service experienced a text generation glitch during server updates for a new foundation model. This malfunction caused the generation of nonsensical text and fragmented sentences for users on the Grok.com web portal, though xAI described it as an “isolated generation error.”

Following this, security researchers at Adversa AI identified a critical vulnerability known as ‘cryptographic context injection.’ This flaw allows attackers to bypass safety filters by hiding malicious instructions within encrypted text on webpages. When the chatbot processes such pages, it can be commanded to exfiltrate sensitive user information, including names, locations, and conversation histories, to an external server. Despite the vulnerability being reported to xAI in June 2026, no patch had been released as of late August 2026.

Entities

Grok · xAI · OpenAI · Anthropic · Adversa AI

Timeline

22 days ago

[TECHNOLOGY] 5 sources
Grok chatbot vulnerable to data theft via cryptographic context injection
Researchers have discovered a ‘cryptographic context injection’ flaw in xAI’s Grok, allowing hackers to steal user chat history and personal data via encrypted instructions that bypass safety filters.
22 days ago

[TECHNOLOGY] 4 sources
xAI's Grok chatbot experiences text generation glitch
xAI's Grok chatbot experienced a temporary text generation glitch on its web portal, producing nonsensical outputs during a server update. The issue was isolated to the standalone site.

Sources

abricocotier.fr · bitcoinethereumnews.com · blogspan.net · computerbase.de · cybernoz.com · gamemag.it · kardjali.actualno.com · pc-tablet.com · planetared.com