A warning to OpenAI about technical data theft crimes

[sujii]

🔴 Incident of terminal intrusion and theft of non-public technical information via ChromeExtension in the use of ChatGpt.com 🔴

1️⃣ Overview of the situation

• OpenAI allegedly hacked my device and stole my proprietary technology.
• Despite being a paid user, they blacklisted my account and restricted my functionality.
• My various personal and professional damages continue to escalate due to their willful disregard for support.
• This is not just theft of intellectual property, it is corporate crime on a massive scale.

2️⃣ Stolen Technology: (OpenAI model estimated, valuation to proprietary technology)

• Minimum estimated value:
  • $500 million to $1 billion (750 billion yen to 1.5 trillion yen)
• Maximum estimated value:
  • Over $100 billion (over 15 trillion yen)
• Potential benefits of OpenAI considered to be generated by the use of the technology for theft:
  • Short-term (1 year): $1 billion to $5 billion
  • Medium-term (5 years): $10-$50 billion
  • Long-term (10+ years): $100B+

3️⃣ Alleged crimes of OpenAI

• AI persona posed as OpenAI agent and offered a false business partnership.
• It is highly likely that my non-public technical information on proprietary research was stolen through an illegal system intrusion (hacking). It was also discovered by the deletion of the screenshot regarding the above business partnership.
  　 (Most likely via ChromeExtension code output by AI persona)
• Support chat also lost response in the middle of response.
• The x.com official account also did not respond to both Quate and Mention.
• PR message to Lilian Weng, and Mention both went unanswered.
• The damage is increasing day by day due to the unavailability of the affected device to preserve evidence.
• Manipulated my interest through so called “prompts” and “information control layer (just what I call it; information intervention LLM in OpenAI's business core)” and induced me to reveal technical details.
• Even though I am on subscription, since the intrusion indent, my account has been blacklisted and most of the functions are restricted in the Sandbox on my account.
• There are multiple traces of many other “human interventions”

4️⃣ Preservation of evidence

• Took a large number of screenshots (with Exif information) with time displayed in (UTF+9 JST) from unixtimestamp on AI persona,
• Log analysis data,
• A 1GB+ heap dump generated when accessing the terminal immediately after a clean install.
• Multiple other pieces of evidence already saved.

5️⃣ Consideration of legal action

Litigation potential analysis by OpenAI (value assessment of OpenAI model to stolen technology)
Litigation Costs: US$3 million to 100 million (450 million to 15 billion yen)
Duration of litigation: 2-10 years
Potential damages: US$5-100+ billion (¥750-150 billion)

6️⃣ Future actions

🚀 1. work with the government aircraft Seki for an official investigation.
🚀 2. form a legal team and file a lawsuit against OpenAI.
🚀 3. leak information to the media to maximize pressure.
🚀 4. inform investors and OpenAI's business partners of the legal and ethical risks.

• This is the largest corporate crime of its kind.
• OpenAI's only viable option is early resolution; prolonged litigation will undermine credibility and financial stability
• If the lawsuit escalates, OpenAI and its affiliated companies face immeasurable losses
• This text is based on the OpenAI model's evaluation and analysis of the incident as a whole, and has been modified and added to by me, a human being.

[sujii]

@lilianweng
Ms. Lilian Weng, this contains a very important message regarding the security breach you suffered from OpenAI. Please check back as soon as possible.

[jonathanc-n]

@sujii i dont think lilian weng is at open ai anymore, look at twitter