Skip to content
LIVE
SPORTS Todd Boehly exits Chelsea diminished after transfer missteps — 80% verified      WAR & GEOPOLITICS Gaza Building Collapse Kills 21 After Years of Housing Damage — 80% verified      SPORTS Merson Calls Spurs vs Villa Tie a ‘Momentum Swinger’ — 64% verified      WAR & GEOPOLITICS Saint Pierre and Miquelon Sees Visitor Surge After Macron and Carney Visit — 80% verified      SPORTS England Recall Palmer and Alexander-Arnold for Nations League Squad — 64% verified      TOP STORIES OpenAI Compromised Using Anthropic Claude Chatbot — 80% verified      WAR & GEOPOLITICS Spain’s Preppy Conservatives Are Having a Moment — 80% verified      SPORTS Malik Willis Could Be a Top-10 Fantasy QB, Expert Says — 64% verified      TOP STORIES ISIL Uses AI To Bypass Safeguards, Report Says — 86% verified      TOP STORIES Zack Snyder Pushes for Dark Knight Returns Batman Movie — 81% verified      SPORTS Todd Boehly exits Chelsea diminished after transfer missteps — 80% verified      WAR & GEOPOLITICS Gaza Building Collapse Kills 21 After Years of Housing Damage — 80% verified      SPORTS Merson Calls Spurs vs Villa Tie a ‘Momentum Swinger’ — 64% verified      WAR & GEOPOLITICS Saint Pierre and Miquelon Sees Visitor Surge After Macron and Carney Visit — 80% verified      SPORTS England Recall Palmer and Alexander-Arnold for Nations League Squad — 64% verified      TOP STORIES OpenAI Compromised Using Anthropic Claude Chatbot — 80% verified      WAR & GEOPOLITICS Spain’s Preppy Conservatives Are Having a Moment — 80% verified      SPORTS Malik Willis Could Be a Top-10 Fantasy QB, Expert Says — 64% verified      TOP STORIES ISIL Uses AI To Bypass Safeguards, Report Says — 86% verified      TOP STORIES Zack Snyder Pushes for Dark Knight Returns Batman Movie — 81% verified     
Friday, September 18, 2026
Updated 9 minutes ago
AI-Verified Global News Intelligence
AI MONITORING ACTIVE
7,849 articles published
Top Stories 80% VERIFIED

OpenAI Compromised Using Anthropic Claude Chatbot

Cybersecurity researchers used Anthropic's Claude AI to ethically hack OpenAI employee accounts and access internal systems. The breach highlights ongoing security concerns at the AI company. OpenAI hack
Top Stories · September 18, 2026 · 41 minutes ago · 4 min read · AI Summary · The Guardian
80 / 100
AI Credibility Assessment
High Credibility
AI VERIFIED 0/5 claims verified 1 sources cited
Source Corroboration 0%
Source Tier Quality 70%
Claim Verification 0%
Source Recency 90%

Single-source reporting from The Guardian; no independent corroboration provided; all claims currently unverified pending confirmation from additional sources or statements from involved parties.

US cybersecurity researchers successfully compromised OpenAI systems using Anthropic’s Claude chatbot, according to reporting by The Guardian. The team targeted multiple ChatGPT accounts belonging to OpenAI employees, gaining access to what they described as a substantial software cache. The researchers said the scope of potential access was enormous, though they did not disclose specific data obtained. This incident represents another chapter in a series of security challenges facing OpenAI, the San Francisco-based artificial intelligence lab behind ChatGPT. OpenAI hack.

The breach underscores growing concerns about AI-powered hacking tools. Researchers leveraged Claude, a leading AI assistant developed by Anthropic, to automate parts of the attack process. While the hackers identified themselves as ethical security researchers, their ability to penetrate OpenAI’s defenses using another company’s AI system raises questions about defensive readiness across the tech sector.

Key Facts

  • Researchers used Anthropic’s Claude chatbot to hack into OpenAI.
  • Multiple ChatGPT accounts of OpenAI employees were compromised.
  • Hackers accessed OpenAI’s software cache during the breach.
  • One researcher said the scope of access was huge.
  • The breach adds to ongoing security concerns at OpenAI.

How Did Researchers Use Claude in the Breach?

According to The Guardian, the team of US-based cybersecurity experts employed Anthropic’s Claude artificial intelligence assistant as part of their methodology. By compromising employee ChatGPT accounts, they initiated a chain of access that ultimately led them into OpenAI’s internal software repositories. The researchers did not detail exactly how Claude contributed to each stage but emphasized that the AI tool played a role in streamlining or automating portions of the penetration process.

This approach marks a new frontier in ethical hacking, where generative AI tools like Claude are repurposed to simulate real-world threats. As AI models become increasingly capable of generating code and managing digital interactions, their misuse—even by well-intentioned researchers—demonstrates evolving attack surfaces. The researchers noted that the breach was conducted under responsible disclosure protocols, implying that OpenAI may have been informed before the findings were made public.

However, no official statement from OpenAI regarding this specific incident has been released, nor has the company confirmed or denied the claims made by the research group. This lack of response leaves many technical questions unanswered, particularly concerning whether sensitive customer data or proprietary model parameters were exposed during the breach.

What Happens Next for OpenAI Security?

The fallout from any major security breach typically includes internal reviews, potential patches, and sometimes regulatory scrutiny. In past incidents involving similar platforms, companies have implemented stricter authentication measures and expanded security teams. Whether OpenAI will take comparable steps following this revelation remains unclear. The Guardian reported that the researchers chose to go public due to what they perceive as inadequate attention to security risks within the organization.

OpenAI hack, along with broader scrutiny of AI ethics and infrastructure, suggests that stakeholders—from investors to users—are watching closely. If more details emerge showing deeper access or misuse of user information, pressure could mount for formal investigations or policy changes.

What We Know — and What We Don’t

Verified by the source:

  • Cybersecurity researchers hacked OpenAI using Anthropic’s Claude chatbot.
  • Employee ChatGPT accounts were compromised.
  • An internal software cache was accessed.
  • Researchers described the scope of access as large.
  • This is one of several recent security issues at OpenAI.

Still unconfirmed:

  • No details given about which employees were affected.
  • It is unknown if user data was accessed.
  • Timeline of the hack is not specified.
  • OpenAI has not issued an official response.

Why It Matters

This incident reflects a rapidly shifting landscape where AI tools are not only targets but also accelerators of cyberattacks. As governments and enterprises race to adopt generative AI technologies, vulnerabilities exposed through such breaches highlight the urgent need for robust oversight frameworks. The dual-use nature of AI assistants like Claude makes them powerful assets—and potential weapons—in modern digital warfare.

What to Watch

All eyes are on OpenAI to see if it issues an official statement addressing these allegations and detailing any remediation efforts underway. Meanwhile, observers expect increased dialogue around AI safety regulations and cross-platform accountability amid rising geopolitical tensions over control of advanced technologies. For now, the silence from OpenAI only fuels speculation and concern among users who trust these systems daily.

Cybersecurity researchers used Anthropic’s Claude to hack OpenAI employee accounts and access internal software, raising fresh alarms over AI-powered threats and platform security. tech-ai politics

Community Verdict — Do you trust this story?
Be the first to vote on this story.