Skip to the story
All StoriesPoliticsConflict & SecurityBusiness & EconomyScience & TechHealthClimateMedia
All stories
Business & Economy241 outlets ran this4 min read

Researchers Hack OpenAI Using Anthropic AI Model

Cybersecurity team exploited a forum vulnerability to access internal code, earning a bug bounty.

AI-assisted coverage comparison, editor-supervised · How this was made

Published
El logotipo de Claude AI, una inteligencia artificial de Anthropic. EP

What this story says

  • Cybersecurity researchers accessed OpenAI's internal systems using Anthropic's Claude AI model.
  • The hack exploited a vulnerability in OpenAI's community forum related to image processing.
  • Researchers gained access to an OpenAI employee's ChatGPT account, linked to the company's code repository.
  • OpenAI awarded the researchers $6,500 for discovering and reporting the vulnerabilities.

Who covered it

Left 25%(20)Centre 41%(33)Right 34%(27)

Percentages are shares of the 80 outlets carrying a published leaning rating. 161 of the 241 outlets we know ran this story carry no rating and are not counted in them. Coverage measured .

Trust

50/100

Craft

65/100

Hype

45/100

241 sources · methodology

Cybersecurity researchers from Hacktron AI successfully accessed internal systems at OpenAI by leveraging Anthropic's Claude AI model. The incident, which occurred in July, involved exploiting a vulnerability in OpenAI's community forum that was related to image processing. This allowed the researchers to gain access to an OpenAI employee's ChatGPT account, which was connected to the company's code repository. The details of the breach were later revealed by The Wall Street Journal. OpenAI acknowledged the findings and rewarded the Hacktron AI team with $6,500 for identifying and reporting the security flaws.

Disagreement on timeline

Tech Radar reported that the entire timeline from vulnerability discovery to repository access took less than 72 hours. Eldiario.es stated the attack began on 23 July and that the researchers revealed the details exclusively to The Wall Street Journal now, without specifying a total duration for the exploit.

What the coverage left out

None of the left-rated reports below mention the specific vulnerability related to the libheif exploit or the ImageMagick conversion process that was detailed in the centre-rated Tech Radar report. None of the right-rated reports below mention the specific vulnerability related to the libheif exploit or the ImageMagick conversion process that was detailed in the centre-rated Tech Radar report.

Tags:#Cybersecurity#OpenAI#Anthropic#ClaudeAI#AI#BugBounty

Coverage settled. We checked this 267 times and stopped on 23 Sept 2026, 14:15. The figures above are what it finished at.

How other outlets pictured it

Which photograph to run is each newsroom’s own choice. The leaning beside a name is that outlet’s published rating, not a claim that the pictures divide along it. Every picture is shown from the outlet’s own server and links to the article it ran in.

Researchers Hack OpenAI Using Anthropic AI Model
Tech RadarCentre

How each side covered it

Our own reading of the reporting listed below, written from the outlets’ articles rather than quoted from them. The reasoning is set out on our methodology page.

Left

20 rated outlets

  • Reports from the left led with the fact that researchers used Anthropic's Claude AI to hack into OpenAI's systems. Both Eldiario.es and Business Insider highlighted that the Hacktron AI team was rewarded $6,500 for their discovery. Eldiario.es also noted that the hack occurred in July and was revealed to The Wall Street Journal, and that the researchers used Claude Opus 5 after initial attempts with Opus 4.8 failed. Business Insider described Hacktron as a 'tiny cybersecurity startup'.

Centre

33 rated outlets

  • Centre-rated reports focused on the breach itself and the speed at which it occurred. Tech Radar stated the hack took less than 72 hours from discovery to repository access, and that the researchers used Claude Opus 5 after initial struggles with Opus 4.8. Quartz noted the three-person Hacktron AI team chained two vulnerabilities to reach OpenAI's private GitHub environment in under 72 hours. Fortune reported that three individuals using Claude hacked into OpenAI and accessed its source code for a $6,500 reward. The Wall Street Journal described the incident as exposing 'growing risks in automated cyber threats'.

Right

27 rated outlets

  • Right-rated reports highlighted that researchers used Claude, a competitor's AI, to breach OpenAI. Zero Hedge stated that three security researchers used Anthropic's Claude to breach OpenAI employee accounts and access private company software. Protothema reported that the hack raised 'new concerns about protecting AI models'. India Today noted that Indian-origin researchers hacked OpenAI using Claude and were paid approximately Rs 6.2 lakh (equivalent to $6,500). Latestly mentioned that researchers used Claude to exploit vulnerabilities in OpenAI's community forum, gaining access to an employee's ChatGPT and Codex environment and linked GitHub repositories, demonstrating the breach with a harmless pull request. Voz.us quoted Hacktron AI's CTO, Mohan Pedhapati, saying, 'I don't think we're as strong as Chinese threat actors. We're just three guys with subscriptions to Claude and Codex.'

Questions about this coverage

How did the left and right cover Researchers Hack OpenAI Using Anthropic AI Model?
Of the 80 outlets on this story carrying a published leaning rating, 25% are rated left, 41% are rated centre, 34% are rated right. Those percentages are shares of the rated outlets, not of every outlet that ran it, which was 241. The sections above set out what each side emphasised, in its own terms.
Is Researchers Hack OpenAI Using Anthropic AI Model left or right?
Neither side dominates it. Of the 80 rated outlets on this story, 25% are rated left, 41% are rated centre, 34% are rated right, and no side holds the 70% this site would want before calling a field one-sided. A story is not left or right in any case; the outlets that carried it are what carry ratings.
Is the coverage of Researchers Hack OpenAI Using Anthropic AI Model biased?
Researchers Hack OpenAI Using Anthropic AI Model is one event reported by 241 outlets, and this page does not rate the story as biased or unbiased. What it publishes is the spread: which outlets ran it, where named rating organisations place each of them on the spectrum, and what each side chose to lead with. A leaning rating describes an outlet's record over time, not this article, and the two should not be run together.
Which outlets covered Researchers Hack OpenAI Using Anthropic AI Model?
241 that we know of, every one of them listed further up this page with a link to its own report and to what we hold on the publisher. Nothing here is a summary of somebody else's summary: the outlets are named so the original reporting can be read.
How did researchers gain access to OpenAI's systems?
Researchers from Hacktron AI exploited a vulnerability in OpenAI's community forum, specifically related to image processing. This allowed them to access an OpenAI employee's ChatGPT account, which was linked to the company's code repository, thereby gaining access to internal systems.
Which AI model was used in the hack?
The researchers used Anthropic's Claude AI model, specifically Claude Opus 5, to carry out the breach. Initial attempts with an earlier version, Claude Opus 4.8, were unsuccessful.
What was the outcome for the researchers?
OpenAI acknowledged the security flaws discovered by the Hacktron AI team and rewarded them with $6,500 as part of their bug bounty program. The company stated it has since patched the vulnerabilities.

Read it at the source

242 outlets, grouped by the leaning a published rating gives them. Every headline links to the original; an underlined outlet name opens our profile of that publisher.

Left

20
Show 12 more

Centre

33
Show 25 more

Right

27
Show 19 more

Not rated

162
Show 154 more

How did this read?

About the coverage, not about the story. We do not ask whether you agree with what happened — we have no honest use for that answer.

More in Business & Economy

All Business & Economy →
US Pushes to Resume Russian Gas Exports Through Ukraine3 min readUS Pushes to Resume Russian Gas Exports Through UkraineRead storyUS Stocks Rise as Oil Prices Drop, Tech Shares Steady3 min readUS Stocks Rise as Oil Prices Drop, Tech Shares SteadyRead storySouth Korea Plans $16 Billion AI and Chip Investment3 min readSouth Korea Plans $16 Billion AI and Chip InvestmentRead story
OpenAI Hacked With Anthropic AI | MediaBias News