The notes and accounts from the FediForum in late September suggest that some of "the people who move the fediverse forward", as the conference promotes itself as platforming, are also acutely interested in moving forward the agenda of Meta.
The forum's notes tell the tale. Though a number of topics, including many of genuine benefit, were touched upon, digging through the sessions turns up a path of breadcrumbs that leads straight back to Palo Alto.
Inspiringly, the forum also paused for a moment of self-reflection, in a session essentially grappling with the question, "Why did we only invite white people to the workshop we organized?" https://fediforum.org/2023-09/session/3-c/
Again, the list above is selective, but piecing the mosaic together reveals a picture for a proposed future-fedi that looks a whole lot like something Mark Zuckerberg could work with.
But the central figure, of course, is the surveillance - and this part of the puzzle is already under construction.
The FediForum dedicated no less than four sessions in support of a plan by the IFTAS thinktank for a realtime centralized "AI" surveillance system for the fediverse.
The centralization scheme is being developed in partnership with an entity called Thorn - a for-profit "AI" surveillance privateer which pretends to be a "for the children" NGO. Thorn is hot news lately due to its blatantly corrupt involvement in the EU Chat Control plot, which would destroy the free internet and online privacy in Europe but create a huge business opportunity for Thorn.
The blocklist system IFTAS proposes is called CARIAD - "Consensus Aggregated Retractable IFTAS Allowlist Denylist".
CARIAD's blocking data will be aggregated from two sources. The first is the Facebook Mafia spider-holed at Stanford, which fabricated the CSAM-scare influence operation that roiled the fedi a few months ago. More on them here: https://kolektiva.social/@ophiocephalic/110772380949893619
The second is "an aggregation of at least ten of the largest ActivityPub service providers"; this would seem to be a sugar pill to win over Mastodon gGmbH and a few other megaservers.
The system itself is somewhat similar to that proposed in the Nivenly FSEP plan which has proven so controversial over the last couple of months; except that, instead of centralizing blocklist control with WelshPixie, CARIAD centralizes control with Meta-linked authoritarian techbros.
Other aspects of the IFTAS surveillance scheme are outlined in the slide deck. They include centralized realtime image and video scanning utilizing Thorn's "AI". Transgressive accounts would be auto-reported to authorities. It should also be noted that Thorn technology employs Amazon's facial recognition algorithms.
As a further comment to this prospect, consider that we are now observing how the moderates currently in power in the so-called United States seem to be gift-wrapping policies (KOSA, the border wall, the criminalization of protest and homelessness) for the reactionary extremists who may well succeed them.
Technologies such as Thorn's should be evaluated in the same light. They may - or may not - only detect CSAM for now. But how will they be repurposed if there are drastic political changes in the US or other "democracies"? What beliefs, convictions, sexual or gender identities will come to be mandated as equivalently deviant? Europol already has some ideas: https://balkaninsight.com/2023/09/29/europol-sought-unlimited-data-access-in-online-child-sexual-abuse-regulation/
The workshop notes also reveal that some of the FediForum sessions concerning the IFTAS system featured multiple participants straight from Meta. There's no need to speculate further. The Zuckerberg entity seeks to impose this surveillance technology on the fediverse before federating.
The September FediForum and the solutionist machinations it platformed provide further detail for our understanding of Facebook's designs on the fediverse. The water is gradually being brought to boil, and it remains to be seen if the frogs of the FediPact 🐸 can leap to action in time.
@ophiocephalic but also, I encourage you to participate in a future fediforum such that you can have a say in the future of the fediverse, there's obviously going to be dissenting voices, that's standard for any massive network.
But to say all these things are funded or backed by zuckerberg is disingenuous at best, imo.
@ophiocephalic Meta representatives were present in many sessions, but that doesn't mean the ideas presented are controlled by Meta. Just like Meta is now interested in SWICG.
That's why it's crucial to fund individuals unaffiliated with big companies who are doing the work & participating in SWICG & FEPs & more.
@thisismissem
Again, you're putting words in my mouth. I didn't say anything is "controlled" by Meta. But their influence on these developments is crystal-clear
@ophiocephalic so far Meta have mostly been observing and not giving input, from what I've seen. They're trying to figure out federated social media and if they can make it work for them, but all the links you've posted haven't had Meta involvement to my knowledge
@ophiocephalic "participants" just mean they were in the chat for the presentation, not that they contributed to the material being presented or the discussion. It was an open forum, all were invited to participate
@thisismissem@ophiocephalic It's not realistic to expect everyone with a personal interest due to their private use of the fediverse to participate in an industry forum before you consider their opinion — based on minutes published by the forum operators themselves — to be valid. To be here posting these seeds of doubt based on what you "think" happened, based on only the meetings you participated in, is not good faith engagement in this discussion.
@tokyo_0@thisismissem
There is also an issue to be raised about FediForum and representation - specifically, who gets invited or platformed to represent. Clearly the bias is in favor of white men with an interest in commercialization. I alluded to that with the barb about the diversity session. However, that's an addressable problem if the organizers actually care about it. The main concern of this thread is the ideas they're presenting on how the fediverse should evolve
It's also an unconference, so literally anyone attending could present and participate. We started out with zero sessions listed, and everyone was free to list a session and something they wanted to talk about.
Yes, the timezone didn't favour asia pacific, but hopefully in the future we will see other events that schedule in that timezone too.
When I first became active on here, people were talking about the potential effects of Meta on the Fediverse. Many of us thought the juggernaut would crush this little continent of conviviality. But there are also those ambitious for the Fediverse, wanting to make it big and damn the consequences and minimise those concerns.
You'd have to either be greedy or naive to believe that linking with Meta, or agreeing with its conditions, are in the Fediverse's best interests. Zuckerberg wants to crush all opposition and gobble up human lives. He is a psychopath. This is scary stuff but at least now it's out in the open.
@ophiocephalic if it's opt-in, is it centralised? Keep in mind the CSAM hash databases are centralised because it's next to impossible to create them legally.
@thisismissem
Again, unless every user of every federated instance has the choice of opting in to having every image they post ingested by a third-party "AI" privateer, it doesn't matter whether you trust your own admin or not
@ophiocephalic no, your admin is saying "every image you post will go there", if you disagree wuth that privacy policy, you can indeed opt for a server that doesn't use one of these services.
That's pretty standard in reviewing a service's ToS/privacy policy for whether or not you wish to use it.
@thisismissem
I again repeat: The system ingests every image of every toot received from another instance that it federates with. One instance utilizing the system exposes the entire fediverse to a centralized surveillance system
@ophiocephalic there are certainly projects or ways of setting up fediverse servers to scan all media (e.g., bucket scanning), but it all depends on the server operator's risk tolerance (we've already seen CSAM used as an instance take-down vector)
@thisismissem
Bottom line is this is a system that would centralize surveillant control of the fedi with an abhorrent, disgraced third-party "AI" privateer. That's what you're arguing for
@ophiocephalic tbh, 99.9% of CSAM tooling works on perceptual hash matching, not by using "AI", there may be newer systems that use AI, but to my knowledge Thorn isn't one of them
@ophiocephalic then contracts need to be drafted to ensure that's not happening.
But the scanning only applies to the instance's local media. And frameworks like GDPR mean that a DPA needs to be in place for any processing of data by third-parties.
@thisismissem
This is a totally unaccountable for-profit entity who was willing to turn the EU into an authoritarian surveillance-superstate so they could gain a new business opportunity. No one should be naive about trusting our community and personal expressions with an entity like that
@ophiocephalic is there an alternative to Thorn that can detect & handle CSAM to prevent continued abuse to victims of CSAM & CSE?
As far as I know there's basically only PhotoDNA which is microsoft or something dubious from google.
I tried contacting the German federal police & the FSM.de for advice on CSAM and they were unwilling to provide any information or processes to follow.
@ophiocephalic this is also something I'm proposing (I've a proposal to merged mastodon's domain_allows and domain_blocks functionality, and introduce "federation requests"
But this work isn't simple and needs funding to do.
@thisismissem
That's good to hear, and the FIRES plan is also very interesting. IMO we need to consider the survival of the fedi to be the struggle to keep it human-centered and -scaled, rather than getting pulled into the growth-at-all-costs pathology, and all its attendant dehumanizing technology
@ophiocephalic@thisismissem thanks for the discussion even if I struggled to follow it all. IMO, I don't think you can construct a set of rules that will suffice. Rather you need a social contract based on a set of shared principles and values, if you trust who you federate with you can keep the rules fuzzy enough to allow instances room to operate in ways that make local sense, without damaging the global system. I don't see how Meta can ever be that.
@BenjaminKlein@thisismissem
Thanks for the comment, and actually, you did get the point that Emelia and I both arrived at where you popped in, by referencing trusting who you federate with. That is the allow-list model. I and others are speculating on approaches to that model which could make it viable. Agreed that Meta could not be a part of that!
@ophiocephalic@BenjaminKlein I'm trying to get work approval to merge allows & blocks in mastodon to have just one domain_filters table which would then unlock a more firewall-like approach & allow for an instance to operate in the middle (approval-based federation with requests to federate)
@ophiocephalic its opt-in, and the sharing of blocklists with IFTAS requires consent of instance admins; its also currently in review, and not actively being used, and there's possibility for change.
(I personally don't think a service pushing blocks is necessarily good, but also understand if admins choose to use such a service)
Add comment