CVE-2025-51865 is an insecure direct object reference (IDOR) vulnerability in the Ai2 Playground web service at playground.allenai.org affecting the LLM chat functionality through 2025-06-03. The issue allows unauthorized access to chat thread data by enumerating thread keys exposed in the URL. The vulnerable condition is the application's failure to properly enforce object-level authorization on thread resources, permitting a user to retrieve information associated with other users' chat threads if a valid thread identifier is supplied.
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.
5 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.