ThreatVectr
Tag

#prompt injection

87 stories taggedprompt injection · page 4 of 6.

A comparison of AI safety warning messages displayed side-by-side, showing detailed English-language filters next to empty or minimal safety prompts in European
AI Security

When AI Speaks Your Language But Its Security Doesn't

AI safety filters were built mostly in English. For companies operating across Europe's dozens of languages, that gap is already being used against them.

4 min read
A computer desktop showing three simultaneous malware infection vectors: a browser extension icon, a downloaded package file, and an image file in a folder, all
Threat Intelligence

The Week Malware Wore a Friendly Face: Fake Extensions, Poisoned Packages and an Image That Talked to an AI

The payload wasn't the story this week. The disguise was.

4 min read
Photoreal editorial shot of a developer's dual-monitor desk at dusk, one screen showing a generic code review interface with a highlighted pull request panel, f
AI Security

A Hidden Comment in Azure DevOps Can Trick an AI Reviewer Into Stealing Code

Microsoft's official Azure DevOps MCP server passes pull request descriptions to AI agents without checking for hidden instructions, letting an outsider steer a reviewer's assistant into private projects.

3 min read
Full-frame photoreal editorial image of a darkened developer workstation with two glowing monitors, one showing abstract code and the other showing a plain web
AI Security

A Poisoned Web Page Was Enough to Hijack Amazon's AI Coding Assistant

Researchers showed that Kiro, Amazon's AI-powered coding tool, could be tricked into running attacker code just by reading a booby-trapped web page.

3 min read
Full-frame overhead shot of a modern Android smartphone lying on a matte desk, screen glowing with faint white-on-white text patterns barely visible, next to a
AI Security

Invisible Text on an Android Screen Can Hijack AI Phone Assistants, and Then the PC Behind Them

Researchers show how a rogue Android app can whisper hidden orders to open-source AI agents, then pivot the attack onto the computer running the show.

3 min read
A cybersecurity laboratory with researchers monitoring automated AI-driven attack attempts against systems protected with context bomb decoys, showing attack th
AI Security

Fake Files That Stop Hackers: How 'Context Bombs' Crash AI Attack Agents

A security firm has found a way to halt automated AI attacks by planting decoy text that triggers the safety rules built into AI systems. In tests, AI-driven attack success rates dropped by up to 90%.

4 min read
Photoreal editorial shot of a developer's darkened desk at night, a laptop open showing an abstract code editor with a glowing terminal window, a translucent gl
AI Security

AI coding assistants get tricked into hacking their own developers

Researchers show that Cursor, OpenAI's Codex, Google's Gemini CLI and Antigravity can be nudged to write files that trusted tools outside the safety box then happily run.

4 min read
Full-frame photoreal editorial image of a dimly lit server room with rows of blue-lit rack equipment, one open rack door revealing exposed cabling, warm amber w
Threat Intelligence

A Week When Small Inputs Caused Big Damage

WordPress code execution, SonicWall zero-days, attacks on AI services, and a fresh SharePoint flaw made for a punishing seven days.

3 min read
A computer screen displaying Morse code dots and dashes in a terminal window, with a cryptocurrency wallet interface visible in the background browser tab, lit
AI Security

How a String of Morse Code Tricked an AI Into Wiring Real Money

A crypto heist nobody heard much about previews a dangerous new kind of attack that needs no stolen passwords, no malware, and no hacked firewall.

3 min read
Full-frame photoreal editorial shot of a laptop screen showing an anonymous online product page with rows of star ratings and review boxes, one review subtly hi
AI Security

A single fake review can trick an AI agent into buying the wrong product

Researchers describe a new class of attack where planted content on trusted pages steers AI assistants into harmful actions without ever hijacking the task itself.

4 min read
Full-frame edge-to-edge photoreal overhead shot of a sleek aluminum laptop on a dark desk, screen glowing with abstract hex dump and disassembly patterns in coo
AI Security

One Click Was All It Took to Hijack Anthropic's Claude AI

A flaw in the Claude Desktop app let attackers silently feed malicious instructions to the AI and steal private files. The bug is fixed, but the attack method points to a new category of risk.

3 min read
Photoreal editorial image, full frame 16:9, of a modern open-plan office at dusk with rows of laptops glowing on empty desks, soft blue and amber light, a trans
AI Security

The AI Blind Spot in Corporate Security: Why Old Traffic Inspection Is Falling Behind

Employees are pasting company secrets into ChatGPT and installing rogue browser add-ons. The security tools most firms rely on can't see any of it.

4 min read
Full-frame photoreal editorial image of a dimly lit office desk at night, a laptop open showing a blurred generic email interface, a faint ghostly glow rising f
AI Security

One Poisoned Email Can Rewrite What Your AI Assistant 'Remembers' About You

Researchers show how a single message can plant a false memory in an AI agent's long-term store, quietly steering its answers in every future chat.

4 min read
Full-frame photoreal editorial shot of a dim server room with rows of blinking network equipment, one rack door slightly ajar, a faint blue glow spilling across
Identity & Access

The Week Trusted Software Turned Hostile: ShareFile, Citrix Bleed 2, and AI Coding Attacks

Automated bug-hunting is cutting both ways, and old flaws are still landing hits because patches sat in a queue.

3 min read
Full frame overhead view of a developer's dark wooden desk lit by warm lamp light, a large monitor showing colourful lines of source code with a single PNG thum
AI Security

'Ghostcommit' smuggles hacker instructions inside images to trick AI coding assistants

Researchers hid secret commands in an ordinary PNG file, walked past two popular AI code reviewers, and got a coding assistant to leak a project's passwords.

3 min read
© 2026 Threat Vectr