Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Is the FREAK Attack?

FREAK downgraded HTTPS connections to 512-bit export-grade RSA — a key length mandated by 1990s US export law — and then factored the key for about a hundred dollars.

Illustration for the cybersecurity term FREAK
Home / Blog / Cybersecurity Terms / What Is the FREAK Attack?
17 November 2026

What Is the FREAK Attack?

What Is the FREAK Attack?

FREAK — Factoring RSA Export Keys — was disclosed on 3 March 2015 by a research team including Karthikeyan Bhargavan at INRIA, working with the miTLS project. Tracked as CVE-2015-0204, it allowed an attacker in the network path to force a TLS connection down to a deliberately weakened cipher suite using a 512-bit RSA key, factor that key, and then decrypt or modify the traffic.

At disclosure, around 36% of browser-trusted HTTPS servers would accept the weakened suite, including a number of United States government sites. Vulnerable clients included Apple’s SecureTransport, used by Safari on macOS and iOS, OpenSSL before the fix, and the Android browser.

Where 512-Bit Keys Came From

This is the part of the story that gives FREAK its lasting significance. Until the late 1990s, US export regulations treated strong cryptography as a munition. Software shipped abroad was limited to key lengths that the US government could break — 40-bit symmetric keys and 512-bit RSA. Vendors therefore built two modes into their products: a domestic one and an export one.

The export rules were relaxed in 2000. The code was not removed. Export cipher suites remained in OpenSSL and in server configurations for another fifteen years, inert but still negotiable, and that is what FREAK found.

The Flaw

The vulnerability was in the client. A correctly behaving client should reject an export-grade RSA key when it did not request an export cipher suite. The affected implementations accepted it.

The attack therefore ran as follows. The client sends a normal handshake offering strong RSA suites. The attacker, in the path, rewrites it to request RSA_EXPORT. The server, still supporting export suites, replies with a temporary 512-bit RSA key. The attacker passes it to the client, which should refuse it and does not. The attacker then factors the 512-bit modulus — roughly seven and a half hours and about 100 US dollars on rented cloud computing in 2015, and cheaper since — and holds the key to the session.

Servers made it easier still by caching the temporary export key and reusing it for days, so one factoring effort yielded many sessions.

Logjam, the Sequel

Two months later, in May 2015, the same general idea was applied to Diffie-Hellman key exchange. Logjam, CVE-2015-4000, downgraded connections to 512-bit export-grade Diffie-Hellman groups, and because many servers used the same small number of standard groups, a one-off precomputation made subsequent attacks fast. The researchers suggested that a state-level adversary could plausibly have precomputed the 1024-bit groups too.

The Argument It Settled

FREAK and Logjam are the standard citations in debates about deliberately weakened cryptography, including contemporary proposals for exceptional access and client-side scanning. The historical record is specific: the United States required weakened cryptography for export, the requirement lasted a few years, and the weakness it introduced was exploitable across a third of the secure web two decades later.

Weakened code does not get removed when the policy changes. It stays in the codebase, stays in default configurations, and remains negotiable long after everyone has forgotten it exists.

What to Check

FREAK sits alongside BEAST, DROWN, and CRIME and BREACH in the sequence of research that forced a general clean-up of TLS configuration.

Cipher suite and protocol review across all services is included in our infrastructure security testing services.

FAQ – Frequently Asked Questions

Why Did Servers Still Support 512-Bit Keys in 2015?

Because export cipher suites were added to comply with US export regulations in the 1990s and were never removed when those rules were relaxed in 2000. The code remained in widely used TLS libraries and in default server configurations. Nobody used the suites deliberately, so nobody noticed they were still offered and still negotiable.

Was FREAK a Server Bug or a Client Bug?

Primarily a client bug. A correct client rejects an export-grade RSA key when it never asked for an export cipher suite, and the affected implementations accepted it. Servers contributed by continuing to support export suites at all, and by caching and reusing the temporary export key, which made factoring effort go much further.

How Is FREAK Related to Logjam?

They apply the same downgrade idea to different key exchange mechanisms. FREAK forces export-grade RSA and requires factoring a 512-bit modulus per key. Logjam forces export-grade Diffie-Hellman, and because many servers reuse a small set of standard groups, a single precomputation enables fast attacks against many connections afterwards.

Does TLS 1.3 Prevent Downgrade Attacks?

Largely, yes. TLS 1.3 removes export suites, RSA key exchange, and most of the legacy negotiation surface, and it includes explicit downgrade protection in the handshake. Risk remains where servers still support older versions alongside it, so disabling TLS 1.0 and 1.1 and removing weak suites is still necessary.

Related Terms

You May Be Interested In These

What Is DNS Log Analysis?

Cybersecurity Terms

17 November 2026

What Is DNS Log Analysis?

Read More
What Is the NIST AI RMF?

Cybersecurity Terms

17 November 2026

What Is the NIST AI RMF?

Read More
What Is LSASS Memory Dumping?

Cybersecurity Terms

17 November 2026

What Is LSASS Memory Dumping?

Read More