ChatGPT Reportedly Exposed Users' Private Data Reportedly Due to Bug

ChatGPT reportedly exposed titles of users' chat histories and users' private payment information to other users reportedly due to a bug, which prompted its temporary shutdown by OpenAI.

Risk classification

• Primary risk domain: 2 Privacy & Security
• Primary risk subdomain: 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information

The incident directly involved the unauthorized exposure of users' private chat histories and sensitive payment information to other active users.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The data leak was caused by a technical bug in the underlying software library, representing a lack of system robustness.

Causal factors

• Entity: Other
• Intent: Unintentional
• Timing: Post-deployment

The privacy leak was caused by an unintentional software bug in an open-source library used by the deployed ChatGPT system.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The report describes ChatGPT as an 'AI chatbot', which falls under the Limited Risk category requiring transparency obligations.

AI system and alleged parties

• AI system: ChatGPT (OpenAI)
• AI purpose: Chatbot; Writing Assistant
• Behaviour type: Assistant
• Alleged developer: OpenAI
• Alleged deployer: OpenAI
• Alleged harmed parties: ChatGPT users

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Substantial, indirect Substantial
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Privacy

Reported: The report explicitly describes privacy violations where chat history titles, first messages of new conversations, and personal payment details of users were exposed to other active users.

Directly caused: The bug directly exposed the first and last names, email addresses, payment addresses, credit card expiration dates, and the last four digits of credit cards for 1.2 percent of active ChatGPT Plus subscribers during a nine-hour window.

Indirectly caused: Users' chat history titles and the first messages of their active chat sessions were visible to other random active users in their sidebar.

Inferred additional harm: It is highly likely that many more users had their private conversation topics exposed than the 1.2 percent of Plus subscribers, potentially revealing sensitive personal or professional information contained in those chat titles.

People affected

• Occurrences reported: 1
• People reportedly harmed: 12000
• People reportedly exposed: 1000000

Potential causes

Technology

• Open-Source Software Bug: A bug in open-source software led to unauthorized data exposure.
• Caching Misconfiguration: The system cached and displayed incorrect user session data.

Data Inputs

• Exposed Chat Titles: Chat titles and first messages were loaded into other sidebars.
• Exposed Payment Details: Names, emails, and partial credit card info were exposed.

Human Factors

• User Navigation Trigger: Users navigating to subscription pages viewed cached data.

Process and Methods

• Delayed Detection Window: The bug was active for nine hours before being taken offline.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, consistent, and detailed information directly from OpenAI regarding the nature of the bug, the specific data exposed (names, emails, partial credit card details, chat titles), the duration of the exposure (nine hours), and the percentage of affected Plus subscribers (1.2%).
• Ambiguities identified: The exact total number of ChatGPT Plus subscribers at the time is not specified, requiring an estimate to quantify the affected population.
• Alternative interpretations: None. The incident is clearly a technical data leak/privacy incident.

A technical bug in an open-source library used by OpenAI temporarily exposed chat history titles and partial payment information of some ChatGPT users. The incident represents a minor data privacy breach with negligible national security implications, as it was unintentional, quickly patched, and did not target critical infrastructure or sovereign functions.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: Global users
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The active bug was patched within nine hours, making this an historical event with long-term data privacy policy implications rather than an active crisis.
• Autonomy: Human-controlled. The AI model acts as a passive assistant, and the data leak was caused by an underlying software bug, not autonomous AI decision-making.
• Novelty: Established threat. Data exposure due to caching or session state bugs in web applications is a well-understood and common software vulnerability.

Impact by dimension

• Physical security: Negligible. The incident was a software data leak and did not involve physical systems, critical infrastructure, or kinetic threats.
• Information security: Minor. Exposure of chat history titles and first messages could potentially leak sensitive user inquiries, but there is no evidence of targeted intelligence compromise or information operations.
• Sovereignty: Negligible. No core government operations, electoral systems, or sovereign decisions were impacted by this technical glitch.
• Economic security: Minor. Exposed partial payment details of approximately 1.2 percent of ChatGPT Plus subscribers, representing a minor commercial data breach rather than strategic economic warfare.
• Societal stability: Negligible. No evidence of mass surveillance, algorithmic oppression, or threats to societal stability resulting from this temporary exposure.