Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
ROBOT showed that a padding oracle first described in 1998 was still present in major TLS products nineteen years later, allowing traffic decryption and signing with a server’s RSA key.

What Is the ROBOT Attack?
ROBOT stands for Return Of Bleichenbacher’s Oracle Threat. Published in December 2017 by Hanno Boeck, Juraj Somorovsky, and Craig Young, it demonstrated that a cryptographic attack first described in 1998 remained exploitable against TLS implementations from several of the largest vendors in the industry. The researchers used it to sign a message with Facebook’s own private key—a demonstration chosen precisely because no further explanation of the impact was required.
In 1998, Daniel Bleichenbacher showed that RSA encryption using PKCS#1 version 1.5 padding was vulnerable to an adaptive chosen-ciphertext attack. The attacker sends many modified ciphertexts to the server and observes whether each is accepted or rejected. Each answer narrows the range in which the original plaintext must lie, and after enough queries—originally hundreds of thousands, later optimised considerably—the plaintext is recovered exactly.
This is a classic oracle attack: the server does not reveal the key, only whether a guess produced valid padding, and that single bit repeated often enough is sufficient.
The standard countermeasure, written into the TLS specification, is to make the server’s behaviour identical regardless of whether the padding was valid. On a padding failure the server must generate a random premaster secret and continue the handshake as normal, so that the handshake fails later, in the same way, with the same timing.
Implementing that countermeasure correctly is harder than it sounds. The server must not differ in its response code, its timing, its TCP behaviour, or anything else an attacker can measure. ROBOT’s authors built a scanner that probed for all of these distinctions and found that many products leaked one of them.
The affected list was not obscure: F5 BIG-IP, Citrix NetScaler, Radware, Cisco ACE, Bouncy Castle, Erlang, WolfSSL, and several others. The researchers estimated that around 27 of the top 100 websites were vulnerable at the time of testing, including Facebook and PayPal. Each required its own vendor patch, and a string of individual CVEs was issued.
Two things, both serious:
The attack does not extract the private key itself, but the practical difference is small when you can use it at will.
Disable RSA key exchange entirely. In TLS terms, remove every cipher suite beginning TLS_RSA_ and offer only ECDHE or DHE suites, which provide forward secrecy and do not transmit an RSA-encrypted premaster secret for an oracle to attack. RSA certificates remain fine—the issue is RSA used for key transport, not RSA used for authentication.
TLS 1.3 removes RSA key exchange from the protocol, which is the most durable form of remediation and another reason to prioritise migration. For anything still on TLS 1.2, patch the affected products and verify the cipher suite configuration with a scanner such as testssl.sh; the ROBOT researchers also published a dedicated detection tool.
ROBOT belongs with POODLE, DROWN, Logjam, and Raccoon in a sequence of attacks against pre-TLS 1.3 design decisions. What distinguishes it is that nothing new was discovered. The mathematics were nineteen years old; what the research contributed was the observation that the mitigation is difficult to implement correctly and that nobody had checked.
The defensive conclusion is that a countermeasure which depends on perfectly uniform behaviour is fragile, and removing the vulnerable construct is better than defending it. That is exactly the approach TLS 1.3 took. For anyone responsible for TLS configuration, the operational version is simpler still: audit cipher suites on every endpoint, not just the public website, and remove options you are only keeping for compatibility with something you have not identified.
Our Network Security Services cover exactly that review across web, mail, VPN, and management interfaces.
No. It gives the attacker the ability to perform private key operations by using the server as an oracle, which allows decryption of recorded sessions and forging signatures. The key itself is never extracted, so it does not need to be rotated for that reason alone, though many organisations chose to reissue certificates as a precaution after confirming exposure.
No, and the distinction matters. RSA certificates remain perfectly acceptable for server authentication. The problem is RSA used for key transport, where the client encrypts the premaster secret with the server’s public key. Configure the server to offer only ECDHE or DHE cipher suites, and the RSA certificate continues to work while the vulnerable construct disappears.
The countermeasure requires the server to behave identically whether padding was valid or not, across response content, timing, error codes, and connection handling. Any measurable difference reconstitutes the oracle. That is unusually demanding to implement and to test, and because the attack was considered solved in 1998, few products were ever checked against it until the ROBOT research did so systematically.
Scan with testssl.sh or sslyze to list the offered cipher suites, and confirm that no TLS_RSA key exchange suites are present. The ROBOT researchers published a dedicated scanner as well. Include mail servers, VPN gateways, load balancers, and device management interfaces, since those are the places where legacy cipher suite configurations most often survive untouched.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.