AIZANOI NEWS

Wednesday, 23 September 2026

Technology

Indian researchers used Anthropic's Claude Opus 5 to break into OpenAI's community forum, exposing cross-vendor AI risk

A three-person team from Indian cybersecurity firm Hacktron disclosed on Wednesday that in late July they used Anthropic's Claude Opus 5 model to access user accounts on OpenAI's community forum, demonstrating how one frontier AI tool can be weaponised against a rival's infrastructure. Researcher Mohan Pedhapati described the technique as a 'force multiplier'. The disclosure lands a day after Anthropic's launch of Claude Opus 5.5 and adds urgency to talks among OpenAI, Anthropic and Google DeepMind about coordinated safety red-teaming and a US-led frontier-model standards body.

By News Desk · Edited by Editorial Desk ·

cybersecurityAnthropicOpenAIClaudeAI red teaming

Sources

Policy Wire