importantSYS.SOURCE: The Hacker News• 2026-08-05T21:06:03+05:30

Unauthorized AI Model Access Service Exposes User Prompts to Operators

#ai security #api security #data privacy #cybercrime #threat intelligence

A service named Poison Claude offers unauthorized discounted access to Anthropic's AI models by routing prompts through compromised accounts, exposing user data to the operator. The service highlights risks in API security and data privacy, with similar gray-market platforms exacerbating concerns over AI model misuse.

Cybersecurity researchers have discovered more than half-a-dozen services advertisements for illegal access to artificial intelligence (AI) models on underground cybercrime forums and messaging platforms.

One such service, Poison Claude, claims to offer access to Anthropic's large language models (LLMs), including Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6.

"Advertisements for Poison Claude

Read original article

*** END OF TRANSMISSION ***

> ZED DELTADB: VERSION CONTROL SYSTEM WITH CONVERSATION-LINKED CHANGES > HOBBY PROGRAMMING COMMUNITIES' RESISTANCE TO LLMS: A CULTURAL AND TECHNICAL ANALYSIS > MICROSOFT'S AI REVENUE HEAVILY DEPENDENT ON OPENAI, DISCLOSURES REVEAL > CASTFORM AND NEON ACHIEVE GPT-5.6 SOL-LEVEL RETRIEVAL ACCURACY WITH 100X COST REDUCTION USING OPEN-SOURCE MODELS > THE IMPACT OF SYCOPHANTIC AI ON PROSOCIAL BEHAVIOR AND USER DEPENDENCY > GOOGLE ANTIGRAVITY 2.0 PLATFORM FOR AGENT-FIRST DEVELOPMENT > REDDIT INTRODUCES AI MODERATION TOOLS TO REDUCE RELIANCE ON KARMA FOR NEW USERS > LUCID MOTORS DELAYS AFFORDABLE EV LAUNCH AMID QUALITY AND STRATEGIC OVERHAUL > ATLASSIAN ROVO AI VULNERABILITY ENABLES ZERO-CLICK DATA EXFILTRATION > EXPLOITATION OF LEGITIMATE CLOUD SERVICES IN PHISHING CAMPAIGNS> ZED DELTADB: VERSION CONTROL SYSTEM WITH CONVERSATION-LINKED CHANGES > HOBBY PROGRAMMING COMMUNITIES' RESISTANCE TO LLMS: A CULTURAL AND TECHNICAL ANALYSIS > MICROSOFT'S AI REVENUE HEAVILY DEPENDENT ON OPENAI, DISCLOSURES REVEAL > CASTFORM AND NEON ACHIEVE GPT-5.6 SOL-LEVEL RETRIEVAL ACCURACY WITH 100X COST REDUCTION USING OPEN-SOURCE MODELS > THE IMPACT OF SYCOPHANTIC AI ON PROSOCIAL BEHAVIOR AND USER DEPENDENCY > GOOGLE ANTIGRAVITY 2.0 PLATFORM FOR AGENT-FIRST DEVELOPMENT > REDDIT INTRODUCES AI MODERATION TOOLS TO REDUCE RELIANCE ON KARMA FOR NEW USERS > LUCID MOTORS DELAYS AFFORDABLE EV LAUNCH AMID QUALITY AND STRATEGIC OVERHAUL > ATLASSIAN ROVO AI VULNERABILITY ENABLES ZERO-CLICK DATA EXFILTRATION > EXPLOITATION OF LEGITIMATE CLOUD SERVICES IN PHISHING CAMPAIGNS