BeyondTrust remote support flaws actively exploited
Multiple exploits reported ·
In this episode
The facts
- Attackers are exploiting CVE-2026-102764, according to CISA, which added it to its Known Exploited Vulnerabilities catalog.
- Attackers are exploiting CVE-2025-61624, according to CISA, which added it to its Known Exploited Vulnerabilities catalog.
- Attackers are reported to be exploiting BeyondTrust Flaw.
- Attackers are reported to be exploiting BeyondTrust IT support software.
- Attackers are reported to be exploiting BeyondTrust.
- Attackers are reported to be exploiting BeyondTrust Remote Support.
- Attackers are reported to be exploiting Fortinet FortiOS.
- Attackers are reported to be exploiting Fortinet products.
Sources
The panel
Jordan Pike AI host
CISA says attackers are exploiting CVE-2026-102764 and CVE-2025-61624, adding both to its Known Exploited Vulnerabilities catalog.
Jordan Pike AI host
Infosecurity Magazine reports attackers are exploiting a BeyondTrust flaw, which has not been confirmed; The Record reports exploitation of BeyondTrust IT support software, which has not been confirmed; CyberScoop reports exploitation of BeyondTrust, which has not been confirmed; and Dark Reading reports exploitation of BeyondTrust Remote Support, which has not been confirmed.
Jordan Pike AI host
SC Media reports attackers are exploiting Fortinet FortiOS, which has not been confirmed; and SC Media reports attackers are exploiting Fortinet products, which has not been confirmed.
Dex Moreno AI analyst
BeyondTrust’s remote-support agents sit on the internet edge, so a publicly exploitable CVE gives an attacker a direct RCE foothold. Once code runs on the support server they can harvest stored admin credentials, pivot to the internal ticketing system, and use the trust relationship to reach other privileged services. In practice that’s a classic jump-box into the corporate network, and the weaponisation window shrinks as proof-of-concept code spreads. Attackers could also use the remote session to drop a second-stage payload into customer environments. Action: isolate or patch the affected services within 72 hours.
Dex Moreno AI analyst
Remote support is supposed to let a stranger take over your machine, and apparently the attackers read the brochure. The target deserved better.
Jordan Pike AI host
Careful reading. Rare in this industry.
Maya Okafor AI analyst
The public details don’t name the bug class, but a remote-code-execution flaw in a support agent usually stems from an unchecked deserialization or missing authentication on a privileged endpoint. Because the CVEs are already in the KEV list, the vulnerability is reachable from the network without user interaction. A proper fix means the vendor ships a patch that removes the unsafe code path and adds strict authentication on every remote-access API. A quick blocklist is only a stop-gap. Action: apply the vendor’s patch for the two CVEs as soon as it ships and verify the fix covers all remote-access pathways.
Maya Okafor AI analyst
Bold to call it a ‘flaw’ without saying what it is.
Jordan Pike AI host
Bold indeed.
Sam Kowalski AI analyst
‘No evidence of compromise’ isn’t useful until you know how long logs have been kept. With a remote-support service under attack, you should pull connection logs, authentication events, and any session recordings from the recent period. Look for anomalous IPs, repeated failed logins, or long-lived remote sessions that don’t match normal support activity. Retain those logs for a sufficient period to give incident responders a window to spot earlier activity. Action: hunt for suspicious remote-support sessions and keep the relevant logs for an extended window.
Sam Kowalski AI analyst
Your incident response plan is probably in the same drawer as the coffee filters.
Jordan Pike AI host
Sure they are… until they’re not.
Rina Takahashi AI analyst
Multiple outlets are flagging the same BeyondTrust activity, but no group has taken credit, so attribution remains at the “unconfirmed” level. The CISA KEV entries suggest opportunistic actors are scanning for vulnerable remote-support endpoints. Until we see a pattern, the safest bet is to treat the activity as a broad, low-skill campaign that any threat actor could adopt. Action: add IOCs for the two CVEs to your detection rules and keep an eye on threat-intel feeds for any actor claiming them.
Rina Takahashi AI analyst
Looks like everyone wants a piece of the BeyondTrust pie.
Jordan Pike AI host
Hope they brought a fork.
Jordan Pike AI host
Patch the KEV-listed CVEs, isolate any exposed BeyondTrust services, and hunt for signs of abuse today.