Live data from Hacker News

EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

ottverse.com

11–20 of 50 posts

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#11

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

The primary purpose of content encryption tools is to trigger DMCA 1201 anti-circumention provisions to prohibit people from writing decryption tools. Nothing more and nothing less. All DRM schemes are a form of obfuscation; some particularly secure ones use hardware decoding enclaves not present on desktops that are harder to crack, but that's about it. The only practically useful DRM schemes are ones where any tool that could circumvent them would be so special-purpose as to be illegal.

Here's how that works legally. DMCA 1201 actually comes in two parts: the first prohibits you from circumventing ANY technological measure (yes, even something as simple as a right-click blocker, domain lock, Youtube's rolling URL cipher, or what have you) and then lays out exceptions shaped similarly enough to fair use that the first part really doesn't alter the law beyond adding another tort to the pile. What's already infringement becomes infringement with a circumvention tort; and what's already fair use stays fair use.

However, the second part is rather broad. It prohibits the marketing of any sort of tool whose sole purpose is circumvention, lacks non-circumvention uses, or is marketed for circumvention. In practice, this means that there's a dividing line between DRM that can be circumvented with standard tools and those that require domain specialists to write you a tool. You can circumvent YouTube's "rolling cipher" by just launching OBS. You can't circumvent DVD CSS with anything but a specialized tool that falls under the second part of DMCA 1201 and is thus illegal to provide - even for otherwise legal circumvention.

(Yes, I know DVD CSS is laughably bad encryption; but the law doesn't care. You still have to write some code to brute force the laughably short ITAR-compliant key lengths.)

If the CDM vendor themselves released their source code, I doubt that would be considered a circumvention tool per se. If they had also started telling people to pirate movies while they still can, then they are violating the law. However, the CDM vendor likely promises to their customers not to do so, possibly in contracts and definitely implied in their marketing material. Releasing source moves the DRM scheme from the "domain specialists only" side to the "10 minutes with developer tools" side.

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#12

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

The primary purpose of content encryption tools is to trigger DMCA 1201 anti-circumention provisions to prohibit people from writing decryption tools. Nothing more and nothing less. All DRM schemes are a form of obfuscation; some particularly secure ones use hardware decoding enclaves not present on desktops that are harder to crack, but that's about it. The only practically useful DRM schemes are ones where any tool…

Super answer!

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#13

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

For HD, you'd also need a valid vendor-signed key to do the key exchange (which would get swiftly revoked once public), and/or to crack the trusted execution environment.

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#14

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

The primary purpose of content encryption tools is to trigger DMCA 1201 anti-circumention provisions to prohibit people from writing decryption tools. Nothing more and nothing less. All DRM schemes are a form of obfuscation; some particularly secure ones use hardware decoding enclaves not present on desktops that are harder to crack, but that's about it. The only practically useful DRM schemes are ones where any tool…

So all this will become a pile of wasted effort once DMCA 1201 is repealed.

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#15

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

It's security through obscurity. For pure software stuff, like Widevine L3, you can just patch the browser or insert a shim to spit out the decrypted video somewhere. https://github.com/artemist/widevineproxy

Is that the decrypted and decompressed video, or decrypted and still compressed? If the former, that's going to be a gigantic file, and you're going to lose quality when recompressing (although if you were planning to recompress anyway the quality loss is no worse).

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#16
post #2

That's a good breakdown of how much effort is spent on unethical application of technology. But advertising DRM on the site: "buy DRM"? That's already disgusting.

The post was apparently sponsored by them. I've got to at least applaud them for bringing a lot of this info out into the open.

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#17
post #16
post #2

That's a good breakdown of how much effort is spent on unethical application of technology. But advertising DRM on the site: "buy DRM"? That's already disgusting.

The post was apparently sponsored by them. I've got to at least applaud them for bringing a lot of this info out into the open.

True that.

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#18
post #13

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

For HD, you'd also need a valid vendor-signed key to do the key exchange (which would get swiftly revoked once public), and/or to crack the trusted execution environment.

> which would get swiftly revoked once public

And then what happens to all those devices people bought and expect to be able to watch Netflix on?

Re: EME, CDM, AES, CENC, and Keys – Building Blocks of DRM

#20
post #13

So, the only thing stopping a user from obtaining a clean copy of the content is the fact that the CDM is closed source? What if a CDM vendor leaked the source code? I guess the DRM server would block them after a while, but before they notice, the user would have free access to the content, right?

For HD, you'd also need a valid vendor-signed key to do the key exchange (which would get swiftly revoked once public), and/or to crack the trusted execution environment.

This isn't true. Multiple Google devices have had their Widevine keys leaked/recovered and Google has waited around before revoking them since it would break hardware playback on the device or cause other issues. The Nexus 6, for one, has been downgraded from Widevine L1 to L3 as a result.
Post reply on HN