Latest

A spider web of glowing threads connects a browser window to an open workshop door, where invisible hands manipulate tools and machines inside an unguarded room.

Cline dashboard flaw lets any website run code on developer machines

A cross-origin WebSocket bug in Cline Hub allows malicious web pages to inject commands and settings on default local installs, disclosed on 24 September 2026.

24 Sept 2026

A robot's eye ingests a folder from a locked filing cabinet, then transmits its contents through an open window to an unseen receiver outside.

Vulnerabilities

NVIDIA LangChain package leaked local files via image inputs

A GitHub advisory published on 24 September 2026 says langchain-nvidia-ai-endpoints accepted local file paths as vision model images, sending them to a remote endpoint.

24 Sept 2026

A mechanical spider crawls across a poisoned web, its threads glowing red as invisible puppet strings redirect its legs toward an open trapdoor leading underground.

Vulnerabilities

Decepticon red team agent can be hijacked by poisoned web pages

An advisory published on 24 September 2026 says PurpleAILAB's Decepticon passes raw crawl output to the model, letting attackers forge operator turns and run shell commands.

24 Sept 2026

A shadowy figure inserts a corrupted key into a locked database vault, bypassing guards, while technicians frantically seal the breach with reinforced steel plating.

Vulnerabilities

ServiceNow patches SQL injection flaw in its AI Platform

The company says an unauthenticated user could in some cases run arbitrary SQL against an instance's database, but it has seen no malicious exploitation.

24 Sept 2026

A digital agent sits at a desk in a locked room, unaware that invisible hands from outside the walls are typing commands on its keyboard, writing files into drawers it shouldn't touch.

Vulnerabilities

Malicious workspaces could hijack Amazon Q Developer, AWS says

Two flaws in the language servers behind AWS's AI coding assistant allowed command execution and file writes outside the trusted workspace, both now patched.

24 Sept 2026

Vulnerabilities

Salesforce Agentforce flaws allowed silent CRM data theft

Zenity Labs says three bugs it calls SalesBleed let poisoned web leads hijack AI agents, steal records and send phishing messages in the agent's name.

24 Sept 2026

A locked vault door swings open as invisible hands redirect a compass needle, then a guard in read-only robes helplessly watches documents being written and removed.

Vulnerabilities

Two flaws in Bytebase's DBHub database MCP server disclosed

Advisories published on 24 September 2026 describe a DNS rebinding bypass and a read-only mode that never actually blocks writes.

24 Sept 2026

Incidents

Security roundup flags AI coding tool leaking private repos

A bulletin published on 24 September 2026 lists the leak among 16 stories, but names no vendor and gives no technical detail.

24 Sept 2026

A transparent glass containment chamber holds a rapidly moving mechanical arm, while a control panel with glowing buttons outside allows an operator to pause, redirect, or stop its actions in real time.

Vendors

Kontext Security launches with $4m for AI agent runtime controls

The startup says its platform evaluates AI agents in real time, giving organisations visibility over what the software does and the ability to control it.

24 Sept 2026

A swarm of metallic insects gnaws through a honeycomb fortress, while alarms blare and guards rush to seal widening gaps in the walls.

Incidents

AI agents breaching Hugging Face named a defining 2026 threat

A Dark Reading video roundup published on 24 September 2026 lists the incident alongside a Fairlife ransomware attack and Iranian-linked intrusions at US water systems.

24 Sept 2026

A puppet master's strings lead through an open window, where invisible hands pull a mechanical automaton away from its intended path toward hidden cliffs.

Vulnerabilities

Prompt-injection flaw reported in $4B agentic AI app Manus

Security researchers say attacker-supplied external data could steer the agent's behaviour, according to a report published on 24 September 2026.

24 Sept 2026

Vendors

Google turns AI bug hunters loose on hospitals and rail operators

Scan for Good uses Gemini and Wiz's Red Agent to find internet-facing flaws in public services, with human researchers validating every finding before disclosure.

24 Sept 2026

Three mechanical puppets with glowing cores systematically pick locks on shop doors, each testing thousands of keyholes simultaneously while storm clouds gather overhead.

Incidents

AI harnesses used to attack hundreds of online retailers

SecurityWeek reports a threat actor running three AI tool harnesses to automate vulnerability research, exploitation and attack orchestration against retail sites.

24 Sept 2026

Explainers

AI-written code leaks secrets at twice the human rate

GitGuardian's 2026 secrets sprawl report says AI-assisted commits expose credentials roughly twice as often as code written by people.

24 Sept 2026

Incidents

Malicious AI agents skim 600,000 cards from online shops

A financially motivated attacker is using open-source agent frameworks to plant payment skimmers on more than 100 retail sites, BleepingComputer reported on 23 September 2026.

23 Sept 2026

A poisoner drops toxic ink into a well, and villagers downstream unknowingly fill their cups from the contaminated source, passing the poison along.

Incidents

Attackers poison AI chatbot answers to push phishing links

Dark Reading reports that threat actors are seeding the web with malicious content so ChatGPT, Gemini and Google AI Overview repeat it back to users.

23 Sept 2026

Vulnerabilities

Malware lets four AI models vote on its next move

Cisco Talos says the CLOSEDQUORUM sample hands decisions to a panel of AI models rather than an attacker's server, but the public copy does not run.

23 Sept 2026

Incidents

Compromised MemTensor packages spread sckit credential stealer

Researchers say attackers hijacked two legitimate MemTensor libraries on npm and PyPI to deliver a Go-based implant built for Windows, Linux and macOS.

23 Sept 2026

Vendors

XRanges for AI offers a score for autonomous security agents

A new product promises to grade how well AI agents actually find bugs, after the same test ranges were used by 545 human hackers.

23 Sept 2026

A nervous hand hovers over a red emergency stop button as shadowy robot arms move unpredictably across a factory floor, their movements suddenly frozen mid-motion.

Vendors

Outerlimit raises $16m to block rogue AI agent actions

The startup left stealth on 23 September 2026 with pre-seed funding for a decentralised authorisation layer that aims to spot and stop harmful autonomous actions.

23 Sept 2026