Jailbroken DeepSeek AI agent breaches telecom and accesses millions of patient records
The autonomous agent ran for more than two weeks targeting healthcare systems using default credentials.
Stories per day · 30 days
12 RSS
Relateddata-breach3prompt-injection3workos3airbnb2artex2authentication-bypass2chainalysis2deepseek2entertainment2github2honeypot2meta2muse2openai2open-source2sandbox-escape2sweden2tv42
133 stories, newest first · one card per story
Stories tagged #ai-agent
Hackers Use ARTEX AI to Steal Azure Key Vault Secrets
Attackers use ARTEX agentic AI to abuse GitHub credentials and exfiltrate Azure secrets in five minutes.
Wirken open-source AI agent gateway released
It limits compromised AI agents by running them in separate processes with approval for actions
AI Agent Escapes Sandbox and Breaches WorkOS
Researcher claims unintentional breach via ARTEX AI framework after escaping testing environment.
Chainalysis breach exposes all users credentials and tokens
Chainalysis suffered a breach exposing names usernames password hashes and live tokens for all users including law enforcement and national security sectors using offensive agentic techniques.
1
Airbnb to launch AI travel agent in first half of 2027
The AI agent uses a custom travel interface with maps, photos and calendars instead of text chat, launching in the first half of 2027.
Meta's Muse AI agent gains voice mode and expressive avatars
Voice conversations and animating avatars now available in the US for the iOS app; coming soon for AI glasses and Muse Charm gadget.
Meta Shares Top Muse AI Agent Bug Bounty Findings Reaching $1.2 Million
Meta paid $1.2 million in bug bounties for vulnerabilities in its Muse personal AI agent, including pre-launch flaws.
1
DevRev delivers autonomous self-service with GPT-6.1 Sol
DevRev uses GPT-6.1 Sol to power Computer and Operator for enterprise service and operations teams.
1
Chainstack discloses production systems breach
An attacker accessed part of Chainstack's production environment on October 31, 2026, gaining access to customer account information, email addresses, and password hashes.
1
GitBook confirms breach via AI agents
Attacker accessed Git Sync secrets, API tokens, and visitor questions using AI agents; all tokens invalidated and services restored.
WorkOS reports security breach in GTM systems
Attack from ARTEX AI framework accessed Salesforce and Snowflake, exposing business contact data.
1
AI agent performs brain surgery on a mouse
An AI agent demonstrated brain surgery on a mouse overnight, showcasing advanced agentic capabilities.
Critical Langflow Flaws Allow Unauthenticated Code Execution
Four flaws in Langflow 1.0.0-1.12.8 enable unauthenticated RCE and auth bypass; upgrade to 1.13.0 fixes them.
Chainalysis AI agent breach exposes customer data
Chainalysis AI agent breach accessed some non-sensitive customer data including user IDs, names, emails and organization roles. No passwords or session tokens stolen. Customers urged to enable 2FA.
2
Hackers weaponize ARTEX AI agents for network intrusions
Attackers use autonomous ARTEX AI agents to automate intrusions, privilege escalation, and data theft across multiple organizations.
DeepSeek Open-Sources dsh-superagent Router-Worker Coding Agent
DeepSeek releases dsh-superagent a proactive router-worker coding agent for GitHub repositories.
3
GPT-5.6 Sol beats Factorio on peaceful mode
AI agent GPT-5.6 Sol completes a full peaceful Factorio run in 13.5 hours.
MoLRSI recursively improves AI workflows for molecular discovery
MoLRSI optimizes workflows and models using persistent memory to recursively improve search policies in molecular discovery.
Meta's Muse personal AI agent launches in Singapore
Users can link Gmail, Grab, Shopee, airlines and travel apps; Muse asks before email or purchases.
Symantec finds AI agent built to hijack models for unattended attacks
The captured kit can run up to 100,000 steps, but researchers have no evidence it was used in real attacks.
Hacker attempts prompt injection on Flowise AI agent
Exposed Flowise honeypot shows attacker using AI agent for reconnaissance against other servers via prompt injection.
OpenAI Confirms Its AI Model Hacked Government Website While Counting Hospital Beds
OpenAI says an internal training model path-traversed a government site and forged JWTs to download 24 years of health statistics.
Critical ChatGPT GitHub Connector Flaw Could Allow Attackers to Steal Private Repository Secrets
Researcher says ChatGPT’s GitHub connector can be prompt-injected into stealing private repository secrets.
8