Interesting angle. Everyone has already pointed out that there are backups basically everywhere, and from an information standpoint, shaving off a day (or whatever) of edits just to get to a known-good point is effectively zero cost. But I wonder what the cost is of the potentially bad data getting baked into those models, and if anyone really cares enough to scrap it.
Canada has always been appealing for people who were set on the United States but taken aback by the political climate. This goes back decades (famously in the AI space, Reagan is why Canada got Hinton instead of the US) but is exacerbated as of late. China also has pretty amazing investment in tech companies and research institutions, but Mandarin doesn't yet enjoy the same widespread adoption of English
Big caveat that I have the perspective of just one US-based former academic.
Why did Meta release theirs? The better question is, why not? If you aren't at the cutting edge and don't have a moat then releasing them is pure reputational upside with zero downside.
The research costs are not free. The businesses need to recoup the cost in some way shape or form, even if in the long term. Seems expensive as an anti-moat to detrench competitors
It's a winner take almost all competition. Baring a moat, if you aren't at the cutting edge you won't be able to recoup the cost regardless. At that point you might as well release it to the public for reputation.
You might even get lucky and someone else does the same. If you manage to learn from their example you might be more competitive in a future round.
For China it's an existential threat. They cannot let US corporations have exclusive control over this. And they are unlikely to catch up. So by tossing open weights model (note: not open source) out there for the public to use, they are destroying the possibility that Dario and SamA can build a monopoly/duopoly.
I think we are now in the era of oligarchies, and oligarchies maintain power by being highwaymen and extracting tolls, in a kind of rentier capitalist structure.
By throwing LLM models out into the commons, China is disrupting the possibility of this taking hold there.
I personally know 3 victims of brutally violent crime. Flock would have detected, but maybe not prevented, two of these cases, where violence occurred in broad, open daylight near main roads and highways. Crimes occurred in left-leaning, anti-police small midwest city. All of the victims were women.
I would encourage anti-Flockers and anti-authority individuals out here to question their motives and make sure that their voices and actions are best aligned with protecting vulnerable individuals (this also includes trafficked illegal immigrants).
Seems like many folks here might be more concerned with preventing hypothetical/theoretical harm, instead of REAL harm (violent crime, trafficking, vehicle theft)
This implies that the harm caused by this broad surveillance technology is "hypothetical/theoretical", when there is long history in this country's government using private companies to launder otherwise illegal surveillance of political activists[1].
And even if you ignore the historical parallels, there are already cases of: officers using Flock systems to stalk dating partners[2][3], immigration enforcement using Flock data to track targets[4], and ICE/CBP bypassing the systems in place that let local jurisdictions choose not to share with federal agencies[5].
I'll acknowledge that there might be some abuses of the use of Flock data by authorities (thanks for sharing citations). I would argue that this is an access control problem: do police departments have broad, unrestricted, unmonitored persistent access to these video feeds? (I oppose this). Is Flock insisting that police departments should have this access?
> Flock would have detected, but maybe not prevented, two of these cases
I'm glad you acknowledge this, because it highlights what has irritated me about the discussion of crime in the last ~6 years. People seem to expect that crime can be prevented. Our criminal justice system and system of civil rights can only intervene after the crime has occurred, which means it won't prevent anything. Maybe I've misread you personally, and I don't mean to put it all on you, but I think people with your position tend to vastly overstate the deterrent factor of proposed interventions.
Further, only reacting to crime and not seeking to "punish" people before a crime has occurred is exactly how our system should work. When reasoning about crime prevention, a large number of people seem to want police to intervene preemptively. Or they want to punish offenders out of proportion to actual crimes, to prevent recidivism that hasn't happened yet. This type of thinking seems to slide pretty quickly into the "pre-crime" concept of dystopian scifi. We called that stuff dystopian for a reason.
In my opinion what we should do instead to prevent crime is to promote social cohesion, in the form of preventing income and wealth disparity, funding a strong social safety net, help for drug addicts and the mentally ill, etc. People who live happier, more stable lives will have less reason to turn to crime.
(I will also note, crime is lower everywhere in America vs. a few decades ago. Violent crime peaked in the mid 1990s. So it is in some sense a misguided endeavor completely, focusing on problems that are relatively unlikely.)
1. I don't think crime can really be prevented per se, but location-based crime can be discouraged and deterred. Having cameras in public, highly visible places, might make violent criminals (especially professional ones) think twice before committing crimes in these places. This potentially creates safe routes for vulnerable individuals (women, children) where they are less likely to be a victim of crime when following these routes. Privacy-minded individuals willing to take additional risk might opt out by driving a different route.
2. I agree on social cohesion, but this seems impossible in USA, which is a country that is/was a melting pot of immigrants from every place in the world. Embracing a national identity seems like the natural solution for creating social cohesion, but nationalism seems unpopular with half of the country (USA invasions don't help the cause). What is your proposed solution?
3.I will also note, crime is lower everywhere in America vs. a few decades ago. Violent crime peaked in the mid 1990s. This is a macro high level generalization; generally true but not everywhere. I am currently located in the midwest and anecdotally saw crime increase in a small city from 2018-2023 (people I know affected were victims of crime,visible increase in homelessness and drifters). Admittedly, crime levels in 2024-2025 in my region seem to be shrinking, but it's too early to determine probable cause.
Gemini AI had this to say about crime levels, agreeing with your opinion but with caveats:
When looking at the broad stretch from the 1990s peak to 2025, the national story is one of a massive, sustained decline. In 1991, the violent crime rate was roughly 758 per 100,000 people; by 2025, that figure is estimated to have dropped by nearly 60%, reaching its lowest levels in nearly 50 years.
However, the "map of violence" has shifted. While the 1990s were defined by high-intensity violence in massive coastal hubs (NYC, LA), the 2020–2025 era has seen crime "decentralize" into the South, the Midwest, and even rural New England.
When we no longer have concentration camps whose victims are located via Flock, then I'll be okay with them. It's not hypothetical, it's happening right now.
They're putting people in pens and then murdering them, and using Flock cameras to round them up.
I don't want this either. Might this be better attributed to politics/administration/policies? So maybe you're opposed to technologies that assist the governement with tracking people? If Flock starts solving missing abducted trafficked persons cases, would this sway your opinion? Or if the data is highly restricted?
I don't care if Flock is used to cure cancer. It's being used to oppress millions of people right now. We need to stop the harm now. It can be reintroduced later when we have a less psychotic administration.
On a more 'logical' note: compare the number of people it theoretically could help, to the number of people non-theoretically being harmed by it right now. The latter number is higher. On a purely emotionless, mathematical point of view, it is a net negative. (And this is a charitable rationalization; using a theoretical future to justify a present harm is illogical)
It sounds like your point is that people should be willing to give up their privacy in return for the chance of detecting (not preventing) violent crimes.
I think it's also disingenuous (or at best, completely naive) to pretend like harm from Flock and other surveillance is hypothetical/theoretical. Here are just 2 recent examples of REAL harm:
I live in Sausalito just north of SF. We have a few cameras on the way into town. Seeing this map actually makes me feel safer. Sure there are hypothetical privacy issues, but for me they're easily outweighed by safety. I don't really get the issue. Ideally this information would be available to law enforcement, but would require a warrant. Is the problem that they can access all of this data without a warrant now?
Based on comments here, I'm thinking that a lot of the Flock opposition is affiliated with the anti-ICE movements. Flock seems to be viewed here as a tool to persecute and hunt down law-abiding government enemies, potentially bypassing civil liberties.
I don't deny that Flock systems could and may have been used in that manner, but that doesn't seem to be its main purpose or use. Presently I'm seeing Flock as a net win for most law abiding persons, and I believe that its use should and can be highly restricted and monitored as a tool to make the country safer. No reason to throw the baby out with the bath water
I mentioned "left-leaning, anti-police" because this particular city attempted to reduce its police force, due to a general distrust in (local?) police authority. It appears that USA police forces might be more violent because they are far more likely to encounter armed criminals (see source below). Thus it makes sense that some communities would prefer to reduce their police force, and might oppose technologies like Flock that increase the chances of police encounters:
Firearm Prevalence: U.S. officers operate in a society with high civilian gun ownership. A 2019 study published in PMC found that the strongest predictor of police being killed on the job—and consequently, their higher rate of using lethal force—is the level of household gun ownership in that state. In this context, U.S. "caution" often manifests as "command presence" to prevent a suspect from reaching for a weapon.
1. Initial access to physical machine, most likely via phishing malware, reckless employees downloading untrusted content, or bad luck.
2. Malware looks for browser cookies, hoping to steal temporary credentials but instead gains persistent creds, which grant Jira access. People re-use passwords; malware tries this password against AdUser and any other systems or other corp user accounts it can find
3. Direct Jira access used to pivot, that custom Jira app is probed for app vulns (likely given design).
So with a better system the malware has to wait an extra couple hours to get the password (by dropping the non-password authentication cookie and making the user log in again), and it can still prod Jira in the meantime. That doesn't strike me as a very big difference. It's an improvement in security but not a big one.
reply