Nyligen visade Google upp sitt nya meddelandeprogrammet Allo och poängterade att Allo kommer att skydda användarnas meddelanden med ”end-to-end” kryptering. Det Google inte berättade är att den funktionen är avslagen som standard.

Internt inom Google verkar det beslutet ha dragit igång en debatt som resultatrade i ett blogg inlägg från Thai Duong, som ingår i Googles säkerhetsteam.

The burning question now is: if incognito mode with end-to-end encryption and disappearing messages is so useful, why isn’t it default in Allo?

I wish it’s the default (because it’s my feature haha :), but even if it is not default all is not lost. I can’t promise anything now, but I’m pushing for a setting where users can opt out of cleartext messaging. Basically with one touch you can tell Allo that you want to ”Always chat in incognito mode going forward,” and from that moment on all your messages will be end-to-end encrypted and auto-deleted. You can still interact with the AI, but only if you explicitly invoke it, so you don’t have to give up everything for your privacy gain.

Blogginlägget raderades kort efter publiceringen och det har tolkats som att Google försökt tysta kritikerna. delar av inlägget ändrades också flera gånger innan det slutligen raderades.

Disclaimer: I’m one of the engineers in charge of the end-to-end encryption feature in Google Allo. This post (and everything else in this blog) is solely my personal opinion, as I don’t speak for my employer.

Ändrades till:

Disclaimer: This post is solely my personal opinion, as someone from outside the team who consulted on security for Allo.

Ändringarna i texten och slutligen att hela inlägget plockades bort tolkas som att Duong utsatts för påtryckningar inom Google.

The changes also suggest that parties at a much higher pay scale than Duong’s are highly resistant to providing the type of end-to-end encryption that’s on by default in messaging apps such as Signal and WhatsApp. Now that Duong has contradicted that bedrock position, it appears these people are trying to distance the researcher from the Allo design team and suppress his highly relevant opinions on security. Given Duong’s track record devising some of the most important crypto attacks of the past five years, including those dubbed BEAST, CRIME, and POODLE, the move is likely to backfire by giving the appearance that a yet-to-be-articulated agenda is eclipsing key concerns and recommendations of a key security engineer.

Ars Technica