More and more, nation-states are leveraging sophisticated cyber influence campaigns and digital propaganda to sway public opinion. Their goal? To decrease trust, increase polarization, and undermine democracies around the world.
In particular, synthetic media is becoming more commonplace thanks to an increase in tools that easily create and disseminate realistic artificial images, videos, and audio. This technology is advancing so quickly that soon anyone will be able to create a synthetic video of anyone saying or doing anything the creator wants. According to Sentinel, there was a 900% year-over-year increase in the proliferation of deepfakes in 2020.
It's up to organizations to protect against these cyber influence operations. But strategies are available for organizations to detect, disrupt, deter, and defend against online propaganda. Read on to learn more.
[...] As technology advances, tools that have traditionally been used in cyberattacks are now being applied to cyber influence operations. Nation-states have also begun collaborating to amplify each other's fake content.
These trends point to a need for greater consumer education on how to accurately identify foreign influence operations and avoid engaging with them. We believe the best way to promote this education is to increase collaboration between the federal government, the private sector, and end users in business and personal contexts.
There are four key ways to ensure the effectiveness of such training and education. First, we must be able to detect foreign cyber influence operations. No individual organization will be able to do this on its own. Instead, we will need the support of academic institutions, nonprofit organizations, and other entities to better analyze and report on cyber influence operations.
Next, defenses must be strengthened to account for the challenges and opportunities that technology has created for the world's democracies — especially when it comes to the disruption of independent journalism, local news, and information accuracy.
Another element in combating this widespread deception is radical transparency. We recommend increasing both the volume and dissemination of geopolitical analysis, reporting, and threat intelligence to better inform effective responses and protection.
Finally, there have to be consequences when nation-states violate international rules. While it often falls on state, local, and federal governments to enforce these penalties, multistakeholder action can be leveraged to strengthen and extend international norms. For example, Microsoft recently signed onto the European Commission's Code of Practice on Disinformation along with more than 30 online businesses to collectively tackle this growing challenge. Governments can build on these norms and laws to advance accountability.
Ultimately, threat actors are only going to continue getting better at evading detection and influencing public opinion. The latest nation-state threats and emerging trends show that threat actors will keep evolving their tactics. However, there are things organizations can do to improve their defenses. We just need to create holistic policies that public and private entities alike can use to combat digital propaganda and protect our collective operations against false narratives.
(Score: 2) by aafcac on Monday March 20 2023, @03:07PM (8 children)
That doesn't work though with machine learning. You'd have to embed a crypto signature into each frame and hope that the general public recognizes a wrong or missing signature as evidence that it's a forgery. All the viewing apps would have to display the information for it to work. The traditional watermarks are trivial to remove with ML in a way that's hard to detect. Embedding it is better, but probably not much, and it has no guarantee of being noticed.
(Score: 2) by PiMuNu on Monday March 20 2023, @04:26PM (1 child)
> The traditional watermarks are trivial to remove with ML
You got it the wrong way around. "Authoritative news source" embeds a watermark in its stuff, posts to server. Client decodes and presents the watermark to the user. If watermark cannot be decoded, the information is considered low quality and should be discounted. Of course, a bad actor can spam whatever junk they want around the web, but if it is not watermarked, it gets put on the junk pile.
> You'd have to embed a crypto signature into each frame
Watermarking is a thing technically. Video codecs, for example, do not work "frame by frame" as you suggest, and adding a rsa key or whatever to each frame is probably not a big deal anyway.
https://en.wikipedia.org/wiki/Digital_watermarking [wikipedia.org]
(Score: 2) by aafcac on Monday March 20 2023, @06:25PM
Depends what kind your dealing with, but if it doesn't include a visible component, good luck with that As far as frame by frame or not, it doesn't much matter as the technology has been around and there are issues. If you use the same keys, then it can be poisoned to not play or be stripped, if you use different ones, the information can be found and stripped.
But in any case, if you're not directly running the stream, there's the potential for either mitm spoofing or only authorized players being allowed.
Really, this is a case of Pandora's box being opened by people too arrogant to see this coming. The equivalent technology does get cracked and it just takes one misplaced root key to reek havoc.
(Score: 2) by krishnoid on Monday March 20 2023, @04:27PM (1 child)
If the digital signature can be extracted and confirmed automatically by the player, like an https certificate chain, isn't that good enough?
(Score: 3, Informative) by aafcac on Monday March 20 2023, @06:17PM
No, at most that solves the question about it being an official source. Videos that are equivalent to the Zapruder film or the tape of Rodney King being beaten by cops wouldn't have that to start. Any video wouldn't have it to start unless it was embedded during capture. Which more or less puts us back where we are. And that has issues as the "official" version might already be tampered with and there will be questions about whether the Israeli copy of the tape is the reliable version or the Palestinian. Both likely would start with unsigned videos and could easily be tampered with prior to being watermarked.
As far as removal goes, if this becomes this important, expect attacks on the entire chain and for governments to not want to sign keys of individuals that are critical of their official line. Removing all the keys and swapping a different set of authorized keys likely would happen before too long. I mean there's a massive amount of money in it from defense contractors at most major powers to break.
(Score: 2) by hendrikboom on Monday March 20 2023, @06:39PM (3 children)
Of course the liars will have their own crypto signatures to make it look honest.
(Score: 2) by PiMuNu on Monday March 20 2023, @07:00PM
Sure. But at least one can see the audit trail and decide whom to trust.
(Score: 0) by Anonymous Coward on Monday March 20 2023, @07:34PM (1 child)
(Score: 0) by Anonymous Coward on Tuesday March 21 2023, @07:29PM
Crypto signatures are FAKE NEWS!11! Now what?