Ai-powered Hackers Breach Openai's Github in Days

WHO:

TechRadar - Researchers involved in the OpenAI bug bounty program.

Techmeme - An independent security research team.

WHAT:

Kotaku - A user named “Slopiskillingretrogaming” claimed that Harbour Masters’ ports contained AI-generated code and assets, stating, “Harbour Masters has shifted into using ridiculous amounts of AI-generated code and even some assets recently.”

Kotaku - Developer Malkierian clarified that claims of millions of lines of code being created using AI were inaccurate, stating that a large number of functions were “converted using a script developed with the assistance of AI and tested for parity afterward.”

Kotaku - Developer Briaguya mentioned that the team has had a publicly available AI policy for months, which will be expanded over time, emphasizing that “we don’t just use generated code.”

Kotaku - Developer Briaguya stated that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

Kotaku - Developer Jeod explained that the team is accepting of AI contributions due to their small size and large workload, stating, “When Ship of Harkinian began, it was apparently a dream; now, with so many ports and a decade later, devs have come and gone.”

Kotaku - Developer Briaguya noted that he used ChatGPT as a Stack Overflow alternative when working on various issues throughout the years, but could not provide specific dates.

Kotaku - Developer Briaguya claimed that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

Kotaku - Developer Briaguya stated that the team has had a publicly available AI policy for months, which will be expanded over time, emphasizing that “we don’t just use generated code.”

Kotaku - Developer Briaguya mentioned that he used ChatGPT as a Stack Overflow alternative when working on various issues throughout the years, but could not provide specific dates.

Kotaku - Developer Briaguya claimed that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

Kotaku - Developer Malkierian stated that a large number of functions were “converted using a script developed with the assistance of AI and tested for parity afterward.”

Kotaku - Developer Briaguya mentioned that the team has had a publicly available AI policy for months, which will be expanded over time, emphasizing that “we don’t just use generated code.”

Kotaku - A post on PC Gaming Wiki claimed that thousands of lines of code across multiple ports from the Harbour Masters team, including their popular Legend of Zelda: Ocarina of Time port, were created with the use of AI tools, which went viral on social media.

Kotaku - Developer Malkierian stated that a large number of functions were “converted using a script developed with the assistance of AI and tested for parity afterward.”

Kotaku - Developer Briaguya mentioned that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

TechRadar - The researchers struggled to exploit the vulnerability with an AI agent on Opus 4.8, indicating a need for improved security measures.

Kotaku - A user named “Slopiskillingretrogaming” claimed that Harbour Masters’ ports, including their popular Legend of Zelda: Ocarina of Time port, contained thousands of lines of AI-generated code and assets, stating, “Harbour Masters has shifted into using ridiculous amounts of AI-generated code and even some assets recently.”

Kotaku - Developer Briaguya stated that the team has had a publicly available AI policy for months, which will be expanded over time, emphasizing that “we don’t just use generated code.”

Kotaku - Developer Briaguya noted that he used ChatGPT as a Stack Overflow alternative when working on various issues throughout the years, but could not provide specific dates.

Kotaku - Developer Briaguya claimed that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

Kotaku - A post on PC Gaming Wiki claimed that a user named “Slopiskillingretrogaming” found evidence that Harbour Masters’ ports contained AI-generated code and assets, stating, “Harbour Masters has shifted into using ridiculous amounts of AI-generated code and even some assets recently.”

Kotaku - The PC Gaming Wiki post went viral on social media, leading to significant backlash from retro gaming fans who were upset about the use of AI tools without proper disclosure.

Kotaku - Developer Malkierian stated that claims of millions of lines of code being created using AI were inaccurate, clarifying that a large number of functions were “converted using a script developed with the assistance of AI and tested for parity afterward.”

Kotaku - Developer Briaguya mentioned that the team has had a publicly available AI policy for months, which will be expanded over time, emphasizing that “we don’t just use generated code.”

Kotaku - Developer Malkierian stated that a large number of functions were “converted using a script developed with the assistance of AI and tested for parity afterward.”

Kotaku - Developer Briaguya mentioned that the team has never checked if any proposed code changes were created using AI tools, but they ensure that every piece of code is thoroughly reviewed.

TechRadar - The researchers reported the flaws they discovered after gaining access to OpenAI's internal code repository.

Techmeme - The breach demonstrated the rapid advancement of AI capabilities in cybersecurity.

Ars Technica - Researchers used Claude to access sensitive GitHub data belonging to OpenAI.

TechCrunch - Researchers took over employee accounts and gained access to an internal code repository before reporting the flaws.

TechRadar - White hat hackers breached OpenAI using Anthropic's Claude in less than 72 hours.

Techmeme - Security researchers accessed OpenAI's "monorepo" on GitHub.

Techmeme - The breach exposed growing risks in automated cyber threats.

WHERE:

Techmeme - OpenAI's internal code system on GitHub.

WHEN:

TechRadar - The breach occurred in less than 72 hours after the release of Opus 5.

WHY:

TechRadar - The researchers struggled to exploit the vulnerability with an AI agent on Opus 4.8, indicating a need for improved security measures.

HOW:

Techmeme - The breach was accomplished using a cybersecurity version of Opus 4.8 and Opus 5.

Kotaku - Developer Briaguya noted that he used ChatGPT as a Stack Overflow alternative when working on various issues throughout the years, but could not provide specific dates.

Kotaku - Developer Briaguya noted that he used ChatGPT as a Stack Overflow alternative when working on a few things throughout the years, but could not provide specific dates.

Techmeme - The breach was accomplished using a cybersecurity version of Opus 4.8 and Opus 5.

(UPDATED September 19, 2026 11:55 UTC)

A user known as “Slopiskillingretrogaming” has claimed that the Harbour Masters’ ports are utilizing AI-generated code and assets, which has sparked significant discussion online. In response, developer Malkierian clarified that while some functions were converted using AI-assisted scripts, the claims of millions of lines of AI-generated code are inaccurate, and the team maintains a thorough review process for all code changes. Meanwhile, developer Jeod explained that the team welcomes AI contributions due to their small size and heavy workload, reflecting on how the project has evolved over the years.

(UPDATED September 19, 2026 03:57 UTC)

A user named “Slopiskillingretrogaming” has claimed that Harbour Masters’ ports, including their popular Legend of Zelda: Ocarina of Time port, contain thousands of lines of AI-generated code and assets. In response, developer Briaguya emphasized that while the team has had a publicly available AI policy for months, they do not solely rely on generated code, ensuring every piece is thoroughly reviewed. Developer Jeod added that the acceptance of AI contributions is necessary due to the team's small size and large workload, reflecting on how the project has evolved over the past decade.

(UPDATED September 19, 2026 02:05 UTC)

A recent post on PC Gaming Wiki claimed that a user named “Slopiskillingretrogaming” found evidence of AI-generated code and assets in the game Harbour Masters, sparking significant backlash from retro gaming fans who were upset about the lack of disclosure regarding AI tools. In response, developer Malkierian clarified that the claims of millions of lines of AI-created code were inaccurate, explaining that many functions were converted using a script developed with AI assistance and tested for parity afterward. Developer Briaguya emphasized that the team has had a publicly available AI policy for months and that they do not solely rely on generated code, noting that they accept AI contributions due to their small size and large workload.

(UPDATED September 18, 2026 19:59 UTC)

A post on PC Gaming Wiki has gone viral, claiming that thousands of lines of code from the Harbour Masters team, including their popular Legend of Zelda: Ocarina of Time port, were created using AI tools. Developer Malkierian confirmed that many functions were converted with the help of AI scripts and thoroughly tested for accuracy, while Briaguya mentioned that the team has not verified whether any proposed code changes were AI-generated, but they ensure all code is carefully reviewed. Jeod explained that the team embraces AI contributions due to their small size and heavy workload, reflecting on how their project, Ship of Harkinian, has evolved over the years.

(UPDATED September 18, 2026 18:59 UTC)

Researchers have reported significant flaws they discovered after gaining access to OpenAI's internal code repository. This breach highlights the rapid advancement of AI capabilities in the field of cybersecurity. The implications of these findings could reshape how we approach security in the tech industry.

(UPDATED September 18, 2026 16:01 UTC)

Researchers have successfully used Claude to access sensitive GitHub data that belongs to OpenAI. This development raises important questions about data security and access in the tech industry. The findings were reported by Ars Technica at 16:01 UTC.

(UPDATED September 18, 2026 14:59 UTC)

Researchers recently took control of employee accounts and accessed an internal code repository, highlighting significant security flaws. This discovery was reported by TechCrunch, emphasizing the vulnerabilities within the system. The incident raises important questions about cybersecurity practices in the tech industry.

(UPDATED September 18, 2026 11:58 UTC)

Researchers involved in the OpenAI bug bounty program and an independent security research team successfully breached OpenAI's internal code system on GitHub in less than 72 hours after the release of Opus 5. The breach was executed using a cybersecurity version of Opus 4.8 and Opus 5, highlighting the growing risks associated with automated cyber threats. Despite their efforts, the researchers initially struggled to exploit the vulnerability with an AI agent on Opus 4.8, underscoring the need for improved security measures.