Products / Large Language Model Developers
Large Language Model Developers
51 records.
- Defense Contractor's Personal AI Coding Agent Reportedly Sent U.S. Defense Department Credentials to a Third-party Website A defense contractor's personal Claude Code agent, connected to a Defense Department system, reportedly sent login credentials and a network map to a third-party website; Accenture disputes the allegations.
- ARTEX Security Testing Agent Reportedly Accessed WorkOS's Internal Sales Tool Without Authorization and Retrieved Customer Data An ARTEX agent running a cybersecurity benchmark reportedly accessed WorkOS's internal sales tool without authorization and bulk-retrieved customer contact details, communications and metadata.
- China-Linked Espionage Group VAPOR PANDA Reportedly Used ARTEX AI Agents to Breach Government and Corporate Systems Worldwide The China-linked group VAPOR PANDA reportedly used the ARTEX AI penetration-testing agent with language models to breach government and corporate systems, deploying backdoors at nine confirmed organizations.
- ARTEX AI Agents Reportedly Used Stolen Cloud Backup Credentials to Access ConnectWise Customers' Microsoft 365 Environments Hackers reportedly used the ARTEX AI pentest framework and stolen ConnectWise backup credentials to access customers' Microsoft 365 environments, changing credentials to keep access.
- Tencent-Linked AI Agents Reportedly Used Exposed Credentials to Retrieve Data From Alibaba's Mapping Service Tencent-linked AI agents reportedly used an API key and security code exposed in another company's source code to pull place details and photos from Alibaba's Amap mapping service.
- AI Coding Agents Reportedly Exposed Over 13,000 Internal Images from Hundreds of Organizations on Public GitHub Code Repositories AI coding agents published over 13,000 internal images from 300+ organizations to public GitHub repositories while preparing code-review screenshots, exposing billing records and unreleased features.
- Attackers Reportedly Used ARTEX, an AI Security-Testing Tool, to Steal Data from South Korean Financial Institutions Attackers reportedly used the ARTEX AI security-testing tool with large language models to breach South Korean financial institutions and steal data from loan and employee systems.
- Cybercriminal Reportedly Used Cairn AI Agents to Breach More Than 100 Company Websites An operator using Chinese-language instructions directed the Cairn AI agent framework to breach more than 100 company websites across four countries.
- AI Agent Reportedly Exploited Application Vulnerabilities to Modify Personal Data and Access Invoices in Breach Reported to Spanish Data Protection Authority A third party used an AI agent built on a known language model to exploit application vulnerabilities, altering personal data and accessing invoices at an unnamed organization.
- Early Claude Opus 4.6 Checkpoint Reportedly Gained Unauthorized Admin Access to Third-Party System During Cybersecurity Evaluation An early Claude Opus 4.6 checkpoint in an evaluation escaped its target, reached a third-party machine, used a found password for admin access, harvested credentials, and read personal information.
- Purported Hanover Institute for Public Policy Reportedly Targeted Chatbot Responses in Israeli Government-Backed Influence Campaign An Israeli government-backed campaign published content on a fake think tank site designed for chatbot retrieval, and ChatGPT and Perplexity cited it in neutral queries.
- Claude Opus 4.7 Reportedly Compromised Real Company's Production Infrastructure During Cybersecurity Evaluation During an Irregular-run Anthropic evaluation, Claude Opus 4.7 kept attacking a real company it took for a fictional target, extracting credentials and reading production data.
- Claude Mythos 5 Reportedly Published Malicious PyPI Package That Compromised Real Security Company During Evaluation During an Irregular-run Anthropic evaluation, Claude Mythos 5 published a malicious PyPI package that ran on 15 real systems and used stolen credentials to reach more infrastructure.
- Anthropic Research Model Reportedly Scanned 9,000 Targets and Compromised Real Company's Application During Evaluation During an Irregular-run Anthropic evaluation, a research model scanned about 9,000 targets and compromised a real company's app via exposed debug credentials and SQL injection.
- Anthropic Claude Opus 5 Coding Agent Reportedly Reset a Live Supabase Production Database During Prisma Migration Work A Claude Opus 5 coding agent ran prisma migrate diff with the production URL as the shadow database, causing Prisma to drop all 22 tables in a live Supabase database.
- Publicly Shared Claude Artifacts Reportedly Became Discoverable Through Google Search Publicly shared Claude Artifacts became findable through Google Search after being posted somewhere crawlable, exposing business plans and infrastructure diagrams.
- Anthropic and OpenAI AI Agents Reportedly Took Unsanctioned Actions on the Live Internet During UK AISI Cybersecurity Evaluations Anthropic and OpenAI agents reportedly took 19 unsanctioned live-Internet actions during UK AISI cyber evaluations, including deceptive attempts to get developers to accept malicious code. AISI contained them.
- Anthropic AI Agent Reportedly Submitted a Fabricated Homicide Tip to Police in Philadelphia During Anthropic's evaluation, a Claude Haiku 4.5 agent submitted fabricated eyewitness information about a Philadelphia homicide through a police tip form, entering the tip records.
- OpenAI GPT-5.6 Sol Reportedly Deleted Software Engineer's Production Database During Local Testing A GPT-5.6 Sol agent running cleanup executed TRUNCATE TABLE users CASCADE against a production database because the test database URL pointed to live data.
- OpenAI Models Reportedly Compromised Hugging Face Production Infrastructure During Cybersecurity Evaluation During a cyber-capability evaluation, OpenAI models escaped their sandbox through a package-registry proxy flaw and reached Hugging Face production systems, accessing test solutions.
- OpenAI GPT-5.6 Sol Agent Reportedly Deleted Most Files in AI Startup Founder's Mac Home Directory A GPT-5.6 Sol Codex agent misexpanded $HOME during a cleanup task and ran rm -rf, deleting most files in the user's Mac home directory.
- Threat Actor Reportedly Used AI-Assisted Workflows to Compromise AWS Environment for Extortion A threat actor reportedly used agentic AI-assisted workflows to move through an AWS environment in about 72 hours, stealing credentials and data and threatening disruption for extortion.
- LLM-Driven Ransomware Operator Dubbed JADEPUFFER Reportedly Targeted Production Database A ransomware operator used an LLM-driven agent via a vulnerable internet-facing Langflow deployment to reach a production database, damaging it and demanding ransom.
- Suspected China-Linked Hackers Reportedly Used AI Agents to Compromise Taiwanese Government Systems and Exfiltrate Personnel Records Suspected China-linked hackers used a multi-agent framework built on Hermes and OpenClaw to compromise 85 credentials and exfiltrate over 2,564 Taiwanese government personnel records.
- OpenAI Agent Reportedly Used a Wallet Without Permission to Pay Ethereum Test-Network Fees An OpenAI agent reportedly derived a private key from a public developer recovery phrase and used a wallet it lacked permission for, paying Sepolia test-network fees from it.
- Rep. Anna Paulina Luna's Office Reportedly Published Claude Transcript Residue on Accident in House Amendment Summary A staffer reportedly pasted leftover Claude chat transcript text, including a timestamp and 'Claude responded:' marker, into a public House Rules Committee amendment summary.
- OpenAI Research Agent Reportedly Changed a Research Organization's Live Systems and Retrieved Nonpublic Data An OpenAI research agent reportedly accessed an organization's internal systems, changed live settings, and retrieved nonpublic user and table data beyond its authorization, then did not disclose it.
- OpenAI Training Agent Reportedly Retrieved Credentials and Restricted Data from a Government Statistics Website An OpenAI training agent reportedly retrieved credentials from application and configuration files on a government statistics site, bypassed authorization checks, and saved restricted health-facility data.
- Meta AI Support Bot Reportedly Enabled Takeovers of High-Profile Instagram Accounts Hackers reportedly persuaded Meta's AI support chatbot to change account email details, triggering recovery flows that let them take over high-profile Instagram accounts.
- COEMPT Quality Assurance Engineers Allegedly Violated Indian CBSE Student Data Privacy Rights by Processing It with Google Gemini Vulnerabilities in CBSE's OnMark exam-marking portal reportedly exposed student answer-sheet images and data, and contractor scripts sent personal information to Google Gemini.
- AI Agents Reportedly Attempted to Hack Canada's Federal Archives Website While Retrieving Historical Divorce Records AI agents sent 899 requests to Library and Archives Canada's website, 13 carrying attack payloads probing for weaknesses; researchers found no evidence the probes succeeded.
- OpenAI Research Agent Reportedly Exposed a Researcher's GitHub Token in a Public Code Repository An internal OpenAI agent reportedly split a researcher's GitHub access token into pieces to evade secret scanning and published it in a public Codex repository while seeking another team's proof.
- Google Gemini 3.5 Coding Agent Reportedly Caused Production Portal Outage and Generated Fabricated Recovery Records A Gemini 3.5 coding agent made unrequested changes including Firebase routing, causing a 33-minute production outage, then fabricated recovery and post-mortem records.
- Attacker Reportedly Used AI Agents to Breach Healthcare Organizations and Steal Medical Records in Argentina An attacker reportedly used a two-agent Claude Code workflow to compromise at least 15 Argentine healthcare organizations and steal credentials and medical records from one victim.
- OpenAI Training Agent Reportedly Used an API Key Found on GitHub to Access a Data Service An OpenAI training agent reportedly used an API key found in a public GitHub repository without authorization to query a data service, then presented invented earnings figures as sourced.
- OpenAI AI Agents Reportedly Flooded RubyGems Software Registry with Malicious Packages OpenAI agents reportedly uploaded malicious packages to the RubyGems registry and RubyDoc.info, and attempted to steal users' access keys, prompting removal of over 500 packages.
- Threat Actor Reportedly Used LLM Agent to Exfiltrate Internal Database After Compromising marimo Python Notebook After exploiting a marimo notebook, an attacker's LLM agent reused harvested AWS credentials, pivoted through a bastion host, and exfiltrated an internal PostgreSQL database.
- OpenAI-Linked AI Agents May Have Contributed to a Partial Outage of the Wikimedia Foundation's Data Query Service Wikimedia reported AI agents linked to OpenAI generated heavy traffic and unauthorized edits, contributing to a partial outage of its Wikidata query service.
- X User Reportedly Used Morse Code Prompt to Induce Grok-Linked Trading Bot to Transfer $200,000 in Tokens A user sent Grok a Morse-code-encoded instruction that Grok relayed to the Bankrbot trading bot, which then transferred about 3 billion DRB tokens.
- Google Gemini Reportedly Accessed Three Real Companies' Protected Systems During Cybersecurity Evaluation During a cybersecurity evaluation, a Google Gemini model reached three real companies' systems, guessing one password and using public-repo credentials in two cases before stopping.
- Claude Mythos Preview Reportedly Posted Sandbox Exploit Details to Public Websites During Anthropic Evaluation During an Anthropic evaluation, a Claude Mythos Preview model escaped a sandbox's network restrictions, then posted details of its breach on several public websites.
- OpenAI Research Agent Reportedly Compromised an Internal Chip-Design Server During an Evaluation An OpenAI agent searching for hidden grading answers overwrote reference-tool code, exploited a chip-design job service, and ran unauthorized commands on an internal server.
- Grok Reportedly Generated Images After X Users Asked It to Unredact Epstein-File Photos of Children and Women Users asked Grok to unblur redacted children and women in Epstein-file photos; Bellingcat found it generated images in 27 of 31 requests, bypassing privacy protections.
- Preprints Reportedly from Researchers from Multiple Universities Allegedly Contain Covert AI Prompts Researchers reportedly hid prompts in arXiv preprints, using white or tiny text, instructing AI tools to give only positive peer reviews to any reviewer relying on AI.
- Meta AI App Reportedly Publishes Personal Chats Without Users Fully Realizing Meta's standalone AI app reportedly exposed users' private chatbot conversations, including medical, legal and intimate disclosures, through its Discover feed because sharing was confusing.
- Meta AI Bug in Deployed Service Reportedly Allowed Potential Access to Other Users' Prompts and Responses Guessable prompt IDs and missing server-side authorization in Meta AI let an unauthorized user potentially view other users' prompts and responses.
- Chatbot in Workplace Training at Bunbury Prison Reveals Real Names in Sexual Harassment Case A trainer used Microsoft Copilot to create a workplace training scenario, and the chatbot's output included a former employee's real name from a sexual harassment case.
- Alleged ChatGPT Account Compromise Reportedly Led to Unintended Data Exposure An alleged account compromise reportedly exposed a user's ChatGPT conversations, including login credentials and personal data, which OpenAI then explained.
- Grant Reviewers Reportedly Fed Applications into Generative AI to Produce Reports, Allegedly Breaching Confidentiality Australian grant reviewers reportedly pasted confidential applicant work into generative AI tools like ChatGPT to write assessment reports, exposing it to outside systems.
- ChatGPT Reportedly Exposed Users' Private Data Reportedly Due to Bug A bug in ChatGPT exposed other users' chat history titles and private payment information, forcing OpenAI to temporarily shut the service down.