CVE-2024-40594: cleartext conversation storage exposed

CISO Take

The ChatGPT macOS app (pre-July 2024) stored all user conversations as plaintext in a filesystem location readable by any other app on the system, bypassing macOS sandbox protections. If employees used the app to discuss sensitive projects, proprietary code, or client data, that content was silently accessible to any co-installed application or malicious script. Ensure all macOS endpoints have updated the app past 2024-07-05, audit ChatGPT usage policies for sensitive data sharing, and consider endpoint DLP rules covering AI tool conversation artifacts.

What is the risk?

CVSS 2.3 understates the organizational privacy impact. While exploitation requires local access with high privileges (limiting remote attack surface), the sandbox opt-out means any app on the same macOS system can trivially read conversation files — no exploitation skill required. The real risk scales with how much sensitive corporate data employees feed into ChatGPT, which in 2024 is substantial. Organizations in regulated industries (finance, healthcare, legal) or with strict data residency obligations face disproportionate compliance exposure from this vulnerability.

How severe is it?

CVSS 3.1

2.3 / 10

EPSS

0.1%

chance of exploitation in 30 days

Higher than 4% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AV Local

AC Low

PR High

UI None

S Unchanged

C Low

I None

A None

What should I do?

6 steps

Immediate: Verify ChatGPT macOS app is updated past 2024-07-05 across all endpoints using MDM (Jamf/Intune).
Locate and review existing plaintext conversation files before the fix (path was ~/Library/Application Support/com.openai.chat/conversations or equivalent) — assess if sensitive data was stored.
Clear conversation history via the app and remove any residual plaintext files from pre-patch versions.
Deploy DLP rules to flag or block exfiltration of files matching ChatGPT conversation patterns from macOS endpoints.
Establish and communicate an acceptable use policy for AI tools that explicitly prohibits sharing sensitive/regulated data.
For high-sensitivity environments, evaluate ChatGPT Enterprise (with data controls) or block the consumer app via network policy.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact partial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Data Leakage Privacy Violation API AML.T0025 - Exfiltration via Cyber Means AML.T0037 - Data from Local System AML.T0057 - LLM Data Leakage

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Art. 10 - Data and Data Governance

ISO 42001

A.6.1 - Information security policies for AI

NIST AI RMF

GOVERN 1.2 - Policies, processes, procedures and practices for AI risk management

OWASP LLM Top 10

LLM06 - Sensitive Information Disclosure

Frequently Asked Questions

What is CVE-2024-40594?

The ChatGPT macOS app (pre-July 2024) stored all user conversations as plaintext in a filesystem location readable by any other app on the system, bypassing macOS sandbox protections. If employees used the app to discuss sensitive projects, proprietary code, or client data, that content was silently accessible to any co-installed application or malicious script. Ensure all macOS endpoints have updated the app past 2024-07-05, audit ChatGPT usage policies for sensitive data sharing, and consider endpoint DLP rules covering AI tool conversation artifacts.

Is CVE-2024-40594 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2024-40594, increasing the risk of exploitation.

How to fix CVE-2024-40594?

1. Immediate: Verify ChatGPT macOS app is updated past 2024-07-05 across all endpoints using MDM (Jamf/Intune). 2. Locate and review existing plaintext conversation files before the fix (path was ~/Library/Application Support/com.openai.chat/conversations or equivalent) — assess if sensitive data was stored. 3. Clear conversation history via the app and remove any residual plaintext files from pre-patch versions. 4. Deploy DLP rules to flag or block exfiltration of files matching ChatGPT conversation patterns from macOS endpoints. 5. Establish and communicate an acceptable use policy for AI tools that explicitly prohibits sharing sensitive/regulated data. 6. For high-sensitivity environments, evaluate ChatGPT Enterprise (with data controls) or block the consumer app via network policy.

What systems are affected by CVE-2024-40594?

This vulnerability affects the following AI/ML architecture patterns: llm api clients, endpoint ai tools, developer workstations.

What is the CVSS score for CVE-2024-40594?

CVE-2024-40594 has a CVSS v3.1 base score of 2.3 (LOW). The EPSS exploitation probability is 0.14%.

What is the AI security impact?

Affected AI Architectures

llm api clientsendpoint ai toolsdeveloper workstations

MITRE ATLAS Techniques

AML.T0025 Exfiltration via Cyber Means

AML.T0037 Data from Local System

AML.T0057 LLM Data Leakage

Compliance Controls Affected

EU AI Act: Art. 10

ISO 42001: A.6.1

NIST AI RMF: GOVERN 1.2

OWASP LLM Top 10: LLM06

What are the technical details?

Original Advisory

The OpenAI ChatGPT app before 2024-07-05 for macOS opts out of the sandbox, and stores conversations in cleartext in a location accessible to other apps.

Exploitation Scenario

An adversary with code execution on a target macOS system — via malicious app, browser exploit, or phishing payload — can enumerate the ChatGPT app's data directory and read all stored conversations in plaintext without any additional privilege escalation. A supply chain attack against a developer's machine could silently exfiltrate months of ChatGPT history (including source code, architecture discussions, API keys pasted for debugging) alongside other artifacts. Insider threat scenario: a disgruntled employee or contractor with standard user access installs a lightweight keylogger-style app that periodically copies the conversation database to an external server — requiring zero AI/ML expertise and no interaction with the ChatGPT app itself.

Weaknesses (CWE)

CWE-312 Cleartext Storage of Sensitive Information

CWE-312 — Cleartext Storage of Sensitive Information: The product stores sensitive information in cleartext within a resource that might be accessible to another control sphere.

[Implementation, System Configuration, Operation] When storing data in the cloud (e.g., S3 buckets, Azure blobs, Google Cloud Storage, etc.), use the provider's controls to encrypt the data at rest. [REF-1297] [REF-1299] [REF-1301]
[Implementation, System Configuration, Operation] In some systems/environments such as cloud, the use of "double encryption" (at both the software and hardware layer) might be required, and the developer might be solely responsible for both layers, instead of shared responsibility with the administrator of the broader system/environment.

Source: MITRE CWE corpus.