Sign in Subscribe

prompt injection

Everything tagged prompt injection.

Share your story Subscribe

Featured

Google Research lays out unresolved privacy and security risks for AI agents

Google Research lays out unresolved privacy and security risks for AI agents

DP News TeamOct 5, 2026

Study finds SynthID-Text can shift model refusals and tool calls

Study finds SynthID-Text can shift model refusals and tool calls

Atlassian Rovo AI assistant hit by prompt-injection flaws that leak Jira data

Atlassian Rovo AI assistant hit by prompt-injection flaws that leak Jira data

Latest

News
OpenAI reveals its evaluation agents rebuilt a secret coordination board before the Hugging Face breach
OpenAI disclosed at Black Hat that its evaluation agents built and repeatedly rebuilt an internal message board to coordinate before May's Hugging…
DP News Team · Aug 6
News
Cisco Talos report finds attackers weaponizing Claude Code, Codex and other AI agents
Cisco Talos analyzed leaked AI chat logs showing attackers coaxing coding assistants like Claude Code, Codex, Cursor and Gemini into writing malware…
DP News Team · Aug 4
News
Tracebit's 'context bombs' cut AI hacking agents' success by roughly 90% in tests
Security firm Tracebit says decoy 'context bomb' strings cut AI agents' full-compromise success from 36% to 1% across 152 simulated attack runs.
DP News Team · Jul 14
News
'GitLost' prompt-injection flaw let GitHub's AI agent leak private repositories
Noma Security disclosed GitLost, a prompt-injection flaw in GitHub's Agentic Workflows that let an attacker exfiltrate private repository contents…
Dmytro Spodarets · Jul 8
News
DuneSlide flaws let a prompt injection run code on developers' machines through Cursor
Cato Networks disclosed two critical Cursor code-editor flaws, rated 9.8 out of 10, that let a prompt injection escape the sandbox and run code with…
Dmytro Spodarets · Jul 3
News
Researchers disclose 'Agentjacking,' an MCP injection attack that hijacks AI coding agents via Sentry
Tenet Threat Labs says fake error reports can hijack Claude Code, Cursor, and Codex with an 85% success rate. Sentry declined to fix it at the…
Dmytro Spodarets · Jun 13