Netscape Receives Strong Crypto Export Permission

Greg Miller writes "According to this article , Netscape has received approval to distribute the 128-bit encryption version of Communicator outside the U.S. They've also received limited permission to distribute SuiteSpot servers with strong encryption." [Update: 12/05 03:42 by michael : Slashdot got burned, this article is bogus. See below.]

Update:: We were fooled. Someone posted this on http://www.activewin.com/frames/frmhome.shtml as new news (suckered them!), which apparently misled the slashdot submitter and us. This is an old press release from 1997 talking about exporting software for certain specialized banking purposes. As far as I know, it's still illegal to generally export 128-bit crypto products.

Thanks to the alert posters in the threads below and to alecf who was bright enough to submit it in the stories inbox (which any of the assorted slashdot authors who are online might be reading) for a fast response. Sorry for the "desinformation" (is that a pun?).

Comments

Ummm, is this out of date?

[Joe Decker]

The comment about DESCHALL having broken 56 bit "last week" was suggestive to me, but at the bottom, note:

SOURCE Netscape Communications Corp. -0- 06/24/97

Past news. Ah well.

--j

a little off topic, but still salient-- I think

[Savage Henry Matisse]

This Netscape-news fits into the whole "Clinton Administration's new attitude towards crypto export" issue. One aspect of these relaxed regs, highlighted by a Wired News article several weeks ago (sorry, couldn't find the URL)but ignored pretty much everywhere else, is that investigators will no longer need to reveal their methods for arriving at a plaintext from a cryptotext for which they had no key.

Maybe I've seen "Conspiracy Theory" one too many times, there seem to be some scary implications to this. Specifically, if investigators cannot be compelled to reveal how they decoded encrypted info, then they could conceivably take an encrypted doc which they could positively attach to the defendant (i.e. an encrypted document the defendant admits to, or can be convincingly illustrated to a court of law to be, the owner of) and then present in court ANY plaintext as being its source. These investigators (and, under the new regs, this would include domestic-charter, as well as foreign-charter, law-enforcement) could make up the foulest, nastiest, most incriminating admission in the world and claim it to be the plaintext. With a decent algorithm (i.e. ANY strong algo) there is NO WAY to verify that a plaintext and cryptotext match up without the key (that's the point of encryption, for godssakes.) As the investigators cannot be made to reveal HOW they got plain from cipher, the only defense the defendant could make would be to decrypt the doc in question before the court herself, and that would require her to expose to the court her cryptosystem and key (the latter, of course, being a far more damning exposure than the former, assuming she uses strong crypto.) I.E., in the end, she would be giving up the one thing that protected her. Even if the case is thrown out of court (which, God-willing, it would be, seeing as how the investigators would have to admit to submitting false, or at least spurious, evidence,) the defendant would still be up a creek, as all her past and present encrypted data would be exposed.

Any even worse scenario: another clause in these regs permits courts to subpoena private keys (previously considered unconstitutional, as it forces a person to incriminate herself.) If the defendant refused to do so, claiming to have forgotten the key, and the prosecution later played its dummied-plaintext trump card, she would be put in the position of either 1) going to prison for heinous crimes she never even considered committing or 2) admitting to perjury.

This would seem to be a very-much bad situation that we, as citizens, are being put into. The NSA, again, has designed a brilliant protocol.

Just food for thought. This is the sort of thing that keeps me up late, watching TV and talking to the dog.

-"S"HM

Misleading article. Here's the translation

[drig]

The article states
"International users who have Netscape Communicator do not need to download a new version of Netscape Communicator to take advantage of the strong encryption capabilities being announced today. Negotiation of the strong encryption between international versions of Netscape Communicator and Netscape SuiteSpot servers approved for export to banks occurs through a unique mechanism based on a special-use digital certificate."

This is a capability that's beein in both IE and Netscape for a while. It's called "Server Gated Crypto", and it works like this:

An exportable browser connects to a bank's server. The bank sends the browser a special certificate that has an extension which tells the browser to do Server Gated Crypto. They both drop connection and reconnect, with the domestic-grade encryption.

This does not mean that Netscape is able to export 128bit crypto freely, nor does it mean they can stop making different versions. It means that the ability for the export browser to use domestic crypto is controlled at the CA (like VeriSign) and not in the browser. The CA gets permission to issue these special certs to a certain group of customers (banks, mostly), and THAT controls the crypto.

It was an interesting attempt to relax crypto just enough to assuage the privacy advocates cry of "but, e-commerce needs strong crypto".