Attacks on the confidentiality and integrity of internet services are carried out, among other things, with vulnerability exploits. They are used by state-directed cyber espionage groups (cf. Attacker Groups) as well as by cybercriminals who offer, for example, Ransomware-as-a-Service to other groups (cf. Cybercrime Groups).
In the current reporting period, the attack attempts recorded at the BSI MADCAT honeypot indicate an unchanged tense threat landscape from attackers who exploit vulnerabilities in web attack surfaces. Not only were the attack numbers above the monthly values of the previous reporting period for almost all months of the current reporting period. The trend also shows a clear direction beyond monthly fluctuations: a further 48 % increase in the current reporting period (cf. Figure 3-2).
Figure 3-4: Exploitation attacks from July 2025 to June 2026 by exploited protocol (Top 5)
shares in per cent of the top 5 protocols
Some slight structural changes were observed. Among the top 5 exploited protocols, port 22 again stood out, which is typically used by the SSH protocol. However, with 37 % of the exploitation attacks observed at the BSI honeypot in the current reporting period (cf. Figure 3-4), a clear decline in the share was recorded (minus 16 percentage points). SSH is used to establish secure connections in insecure networks such as the Internet, and is therefore used, for example, for account logins. Typical attacks against this port include, for example, brute-force attacks. Here, attackers automatically try out lists of common, weak, widespread, or even calculated usernames and passwords on an SSH port randomly found on the Internet. Poorly protected accounts or misconfigured servers are easily victimized by such attacks, and the increasing performance of AI attack tools will also accelerate brute-force attacks and make them increasingly successful. With RDP and VNC, two remote maintenance protocols are again in the top 5, which again underscores that particular caution is required when exposing services that are primarily designed for use in protected network areas (cf. Figure 3-4).
React2Shell
The exploitation of the vulnerability React2Shell, also known as CVE-2025-55182 (also known under CVE-2025-66478), which was rated with the highest CVSS score of 10.0, was exemplary for the execution of exploitation attacks.
Figure 3-5: Average exploitation attacks on the React2Shell vulnerability
number per sensor
The vulnerability enables attackers to take over server-side processes with the help of exploits in the React server components. Thus, manipulated data containing the attackers' commands can be inserted, which then get executed. Waves of attacks began just a few hours after the disclosure of a proof of concept on December 4, 2025, which confirmed the exploitability of the vulnerability (cf. Figure 3-5). This was followed by a phase of intensified exploitation. The curve then flattened out, but the threat continues.
The security relevance of such a vulnerability is high, as it enables so-called remote code execution without authentication. As a result, the BSI continued to work, among other things through warnings, to ensure that operators check and secure their systems.