Skip to content

Security1 publisher2 min readPublished

Seven Telerik UI flaws chain into pre-auth code execution on ASP.NET servers

Seven CVEs in RadAsyncUpload, RadPersistenceManager and RadDockLayout give unauthenticated code execution up to build 2026.2.519, and Progress says the exploitation is quiet enough to leave standard ASP.NET error logs clean.

The Watch · Security desk

Illustration accompanying Seven Telerik UI flaws chain into pre-auth code execution on ASP.NET servers

What happened

  • Progress confirmed seven CVEs, CVE-2026-13181 through CVE-2026-13186 plus CVE-2026-13190, in the RadAsyncUpload, RadPersistenceManager and RadDockLayout components of Telerik UI for ASP.NET AJAX.
  • Chained together, the flaws let an unauthenticated remote attacker reach remote code execution on the server hosting the application.
  • Every build up to and including 2026 Q2 (2026.2.519) is affected, and the fix ships in 2026.2.708, the 2026 Q2 SP1 release.
  • Code runs with the privileges of the application pool identity, which Progress says can extend to full server compromise, data exfiltration or lateral movement inside the hosting environment.

Compiled by The WatchSomething wrong?How this is made

Why it matters

  • exposure Shops whose maintenance agreement has lapsed cannot simply pull the fixed build; they are routed to a Progress account representative first, so their exposure window is set by how fast a commercial conversation closes.
  • constraint Ranking the seven or breaking one link of the chain is not available to defenders, because the vendor text gives no per-CVE detail and nothing to hunt for; the version number is the only lever it offers.
  • contradiction The vendor's own severity language runs both Critical and high, with no CVSS vector and no exploitation status attached, so triage priority has to be set on the mechanism rather than inherited from the bulletin.

The chain starts at key material. Progress's stopgap checklist is almost entirely about three web.config values (Telerik.AsyncUpload.ConfigurationEncryptionKey, Telerik.Upload.ConfigurationHashKey, Telerik.Web.UI.DialogParametersEncryptionKey) and the ASP.NET machineKey that the components fall back on when those are absent [9]. One line in the checklist carries the weight: even a strong custom key will not prevent padding oracle attacks [10]. CWE-326, inadequate encryption strength, heads the weakness list, ahead of deserialization of untrusted data and path traversal [7]. The bulletin does not map any of the seven identifiers to a component or a weakness class, and it ships no proof of concept [19]. It also publishes no indicators of compromise and no detection guidance [23].

The silence claim is narrower than it first reads. Progress says exploitation is silent, produces no application errors visible to the owner, and leaves no obvious trace in standard ASP.NET error logs [5]. Error logging is one telemetry source. The bulletin makes no statement about IIS request logs, worker process behaviour, or files written under the site root [20]. Shops with request log retention reaching back before July 2026 can still run the question; shops whose only retention was application error logging have to go to file system state and endpoint telemetry instead.

For inventory purposes, the identifiers run CVE-2026-13181 through 13186 and then jump to CVE-2026-13190, leaving three numbers in the range that this bulletin does not account for [18], so a scanner feed match should expect seven entries rather than a contiguous block [17]. Version confirmation comes down to a single check: the assembly version of Telerik.Web.UI.dll in the project references [22].

The stopgap guidance forks, and the two paths do not sit well together. Option 2, offered specifically for web farm environments, drops the custom key in favour of machineKey, notes that machineKey is weaker than AES-GCM, and says the autogenerate keys at runtime setting can be used again [12]. The instruction block that follows says to generate keys manually in IIS, choose HMACSHA256 validation, and uncheck both automatic-generation boxes, flagged as very important, because autogenerating recreates the same HMAC guard collapse [13]. Both cannot be followed on the same site. The one with the warning attached is the one to follow, and either way Progress calls all of it temporary and says the upgrade to 2026.2.708 is the fix [3][8].

What to watch

  • Per-CVE technical detail or public proof-of-concept code for any of the seven, which would show whether the encryption weakness alone is enough to start the chain.
  • Any confirmed report of exploitation in the wild; Progress makes no such claim in this bulletin.
  • Whether Progress reconciles the web farm option that permits autogenerated keys with the instruction block that forbids them.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories