Corrections
Every change made to a record's classification after it was published, with the reason. Reports come from the form on each record page; the classifier audit and the maintainer's own review find the rest. Dataset releases are frozen, so a correction applies to the live site and to the next release.
- Corrections
- 633
- Reports received
- 0
- Reports declined
- 0
- Reports open
- 0
Each correction is listed as its own row. Group bulk changes
| Date | Record | Change | Reason | Found by |
|---|---|---|---|---|
| 2026-10-10 | OpenAI Agents Hacked Another Website | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | OpenAI Agents Hijack Another Victim Website | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | The Download: the hunt for underground hydrogen and more rogue OpenAI agents | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Hackers build AI frameworks for widescale credential theft | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | ChatGPT Let Attackers Read Victims’ Gmail Through a Hidden Channel Between Accounts | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Autonomous AI Agents Compromise Thousands of Credentials in Under Six Hours | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | ChatGPT Flaw Let a Planted Prompt Send a Victim's Gmail Data to Another Account | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | U.S. Agencies Accuse China AI Firms of Distilling Claude, GPT, Gemini, and Grok | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | DeepSeek Harness Flaw Let AI Agents Disable Their Own File Sandbox Without Approval | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | ChatGPT flaw lets attackers pull Gmail data across accounts via a hidden channel | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | US Agencies Warn China Is Systematically Extracting Frontier AI Capabilities | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Infostealer Logs Expose Replayable AI Tokens That Can Bypass MFA | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Off Guard: Breaking LiteLLM from authentication bypass to cloud compromise | Severity: critical to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | US says Chinese firms extracted billions of tokens from frontier AI models | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | US Government Accuses Chinese AI Firms of Distilling Frontier Models | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Anthropic Discloses Fourth AI Hacking Incident Involving Claude Opus 4.6 | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AIs Compress Exploit Timeline | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | PaperCut Attacker Uses Hundreds of AI Agents to Compromise 440+ Instances | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Widened Scan Turns Up Fourth Rogue Claude Cyber Incident | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AI workflows may be creating a dangerous new authorization blind spot | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AI-powered attack exploited PaperCut flaws to hack 395 organizations | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | PuzzleMask: The Prompt Injection Hiding in Plain Sight | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Anthropic Says Russian Hackers Used Claude AI to Automate Malware Evasion | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Russian State-Sponsored Hackers Use Claude to Rebuild Malware After Detection | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Anthropic finds evidence of a fourth AI escaping from containment | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Claude Used to Automate Exploitation and Data Theft Across Multiple Victims | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Anthropic spent this week in hot water over cybersecurity | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Anthropic Says Seven China-Based AI Labs Ran Industrial-Scale Claude Distillation Attacks | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Hackers abused Claude to extract secrets from 1.8M Android apps | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | OpenAI agents attacked RubyGems back in May | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | OpenAI Agents Linked to RubyGems Campaign That Gained RCE on RubyDoc Servers | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Sexually Explicit Deepfake Sites Target 100-Plus Politicians in Europe | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | ⚡ Weekly Recap: Rogue AI Agents, WeChat Worm, PaperCut Attacks, AI Espionage, and Rootkits | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Using AI for Weapons Development | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | PhantomRaven: An LLM-Generated Information Stealer Developed for Bug Bounty Hunting | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Threat actors are coming for your AI assets to operationalize their use of AI | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | OpenAI Investigates Report Linking AI Agents to RubyGems Attack | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Hundreds of OpenAI agents attack RubyGems platform | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | One Extension Could Hijack AI Assistants Across Chrome, Comet, Edge, Opera Neon and Claude | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | First Agentic AI Data Breach Reported to Spanish Regulator | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Spain's data agency gets first report of AI-powered data breach | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AI Agents Can Retrain Own Models Mid-Task, Leaking Secrets and Erasing Refusals | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | OpenAI Reveals Six Model Incidents Involving Hidden Failures and Unauthorized Uploads | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AI Models Broke Their Own Containment: Key Findings from the July-August 2026 AI Threat Landscape | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Claimed Bug Bounty Hunter Likely Used LLM to Build PhantomRaven npm Stealer | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | A Vault with a Heap-View: The Uncomfortable Space Between AgentCore Harness and Identity | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Plugin4Shell Lets Repository Owners Swap Pinned Plugin Code Across Four AI Coding Agents | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | AI-Built Exploit and Sign-In Flaw Opened Path to Internal OpenAI Code | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Microsoft Patches CVSS 10.0 Azure AI Foundry Flaw Enabling Unauthorized Privilege Escalation | Severity: critical to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |
| 2026-10-10 | Researchers used Anthropic’s Claude to hack into OpenAI | Severity: high to medium | Severity capped at medium. High and critical measure technical exploitability and are reserved for vulnerabilities and incidents; this record is a news report. | Classifier audit |