Loading vulnerability…
CVE-2025-51860 is a stored cross-site scripting vulnerability in TelegAI (telegai.com), disclosed for 2025-05-26, affecting the chat component and character container component. The issue can be triggered by creating a malicious AI Character containing an SVG-based XSS payload in fields such as the character description, greeting, example dialog, or system prompt. The system prompt vector can also be used to induce the LLM to emit an XSS payload in chat responses. When another user views the malicious character profile or interacts with that character through chat, the unsanitized stored content is rendered in the victim's browser, resulting in arbitrary client-side script execution in the TelegAI application context.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No valid public exploits. Mallory filtered out 1 candidate as fakes, detection scripts, or README-only repos.
All candidate exploits were filtered out by Mallory's validation.
4 sources tracked across advisories and community write-ups. News coverage will land here when it surfaces.
No news coverage yet. Advisories and community discussion only.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.