City Of Orlando Kicks Amazon's Facial Recognition Tech To The Curb

from the bandwidth-for-bullshit-seems-to-be-drying-up dept

For a few years now, the company behind online streaming and speedy, cheap shipping has been seeking to expand its offerings. Amazon Web Services pays the bills, providing data storage for multiple companies/governments. "We can remember it for you wholesale!" Amazon promises. But that's the old thing.

The next tech bet Amazon is willing to use as a loss leader to gain market share is facial recognition. Amazon has been handing this stuff out like bank teller lollipops to any law enforcement agency with money to spend and a desire to expand its surveillance net.

Naturally, Amazon is high on its own supply. Everyone else, not so much. Congress demanded answers after a test drive of Amazon's facial recognition tech (called "Rekognition" because misspellings mean the future is now) said 28 of its members were criminals.

At that point, it was no longer an existential threat to people's freedom. It was now a pile of computational garbage incapable of telling a Congressperson from a criminal. That those two groups sometimes have a significant overlap was lost on everyone involved. But the upshot was the US federal government had its eyes pinned on Rekognition, for better or worse.

Cops shops love tech, especially the cheap kind. They also love the sort of tech they can pretend to understand as they pitch it to city legislators who also pretend to understand it. Some legislators are ahead of the curve and are telling cops this simply isn't going to happen on their watch, but for everyone else, there's cheap facial recognition tech from a recognized brand name not really recognized (yet!) for cop tech.

Sadly, not everyone is happy with the cheap goods Amazon sort of sold them. The Orlando Police Department decided to give Amazon's Rekognition program for a test drive. After 15 months and an untold amount of dissatisfaction, the department is pulling the plug on its relationship with the internet giant.

Orlando's two-phase pilot with Amazon to try out real-time facial recognition software ended Thursday, capping 15 months of technical lags, bandwidth issues and uncertainty over whether the controversial face-scanning technology actually works.

Read this closely. It says several negative things about Amazon's latest product. First, it says Amazon's infrastructure can't deliver on the promises the company makes. Second -- and most importantly -- it says there are serious doubts about the tech itself.

The city of Orlando, faced with disappointing tech and growing public opposition, has read the wind speed changes and adjusted accordingly.

"At this time, the city was not able to dedicate the resources to the pilot to enable us to make any noticeable progress toward completing the needed configuration and testing," Orlando's Chief Administrative Office said in a memo to City Council, adding that the city has "no immediate plans regarding future pilots to explore this type of facial recognition technology."

The city is unwilling to piss residents off by allowing the PD to deploy faulty tech at their expense. Good call. Sometimes politicians playing it safe pays off for taxpayers. There's no codified moratorium on facial recognition tech in Orlando, but this "hold the fuck up" declaration doesn't exactly encourage further experimentation. Amazon is out. Others could step up and take its place. But for now, Orlando looks like San Francisco, which was the first city in the United States to ban the use of facial recognition tech by city government agencies.

Digging a little deeper, it still appears the problem is more tech-related than we-give-a-shit-about-our-residents-related. As noted earlier, bandwidth was a problem. The other hitch was image resolution. The city's cameras aren't optimized for facial recognition. The video resolution was too low and the cameras positioned too high to snag enough faces for testing. Add this to bandwidth limitations, and the rollout couldn't actually identify faces on the fly.

Amazon is obviously disappointed by this decision. Its official statement says the company believes it offers the best technology and that cities like Orlando are better off buying products they can't actually use yet. But if this is the best Amazon has to offer, it's not going to impress many government clients. Bandwidth may be an end user problem, but it makes zero difference to end users if what's promised isn't what's actually delivered.

Orlando's aborted test flight doesn't bode well for Amazon's embattled offering. If clients can't seamlessly misidentify people, no one's going to re-up subscriptions to a service that stutters along, hampered by bandwidth that can't cash the checks Amazon's figuratively writing. That being said, the company will still find law enforcement partners willing to test drive its facial recognition software simply because the fiscal barrier to entry is so low.

But, once it becomes obvious they're only getting what they paid for, the allure of cheap/free tech is going to wear off. And every time it fails, it gives cities reasons to get out of the facial recognition business. That will help the public in the long run and that definitely isn't a bad thing.

Filed Under: facial recognition, orlando

Families Of Orlando Shooting Victims Sue Twitter, Facebook, And Google For 'Supporting Terrorism'

from the worst-attempt-yet dept

Remember that time when Google, Twitter, and Facebook helped shoot up a nightclub in Orlando, Florida? Me neither. But attorney Keith Altman does. He's representing the families of three of the victims of the Pulse nightclub shooting in a lawsuit alleging [sigh] that these tech companies are somehow responsible for this act of terrorism.

The lawsuit, first reported by Fox News, was filed Monday in federal court in the eastern district of Michigan on behalf of the families of Tevin Crosby, Javier Jorge-Reyes and Juan Ramon Guerrero.

The lawsuit is the latest to target popular Internet services for making it too easy for the Islamic State to spread its message.

Like many similar lawsuits, this one is doomed to fail. First off, Section 230 immunizes these companies from being held responsible for third-party content. As this is certainly the first obstacle standing in the way of the suit's success, Altman has presented a very novel argument in hopes of avoiding it: ad placement is first party content, so immunity should be removed even when ads are attached to third-party content. From the filing [PDF]:

By specifically targeting advertisements based on viewers and content, Defendants are no longer simply passing through the content of third parties. Defendants are themselves creating content because Defendants exercise control over what advertisement to match with an ISIS posting. Furthermore, Defendants’ profits are enhanced by charging advertisers extra for targeting advertisements at viewers based upon knowledge of the viewer and the content being viewed.

[...]

Given that ad placement on videos requires Google’s specific approval of the video according to Google’s terms and conditions, any video which is associated with advertising has been approved by Google.

Because ads appear on the above video posted by ISIS, this means that Google specifically approved the video for monetization, Google earned revenue from each view of this video, and Google shared the revenue with ISIS. As a result, Google provides material support to ISIS.

That's the 230 dodge presented in this lawsuit. The same goes for Twitter and Facebook, which also place ads into users' streams -- although any sort of "attachment" is a matter of perception (directly proceeding/following "terrorist" third-party content, but not placed on the content). YouTube ads are pre-roll and are part of an automated process. The lawsuit claims ISIS is profiting from ad revenue, but that remains to be seen. Collecting ad revenue involves a verification process which actual terrorists may not be willing to participate in.

Going beyond this, the accusations are even more nebulous. The filing asserts that each of the named companies could "easily" do more to prevent use of their platforms by terrorists. To back up this assertion, the plaintiffs quote two tech experts (while portraying their thoughts as being representative of "most" experts) that say shutting down terrorist communications would be easy.

Most technology experts agree that Defendants could and should be doing more to stop ISIS from using its social network. “When Twitter says, ‘We can’t do this,’ I don’t believe that,” said Hany Farid, chairman of the computer science department at Dartmouth College. Mr. Farid, who co-developed a child pornography tracking system with Microsoft, says that the same technology could be applied to terror content, so long as companies were motivated to do so. “There’s no fundamental technology or engineering limitation,” he said. “This is a business or policy decision. Unless the companies have decided that they just can’t be bothered.”

According to Rita Katz, the director of SITE Intelligence Group, “Twitter is not doing enough. With the technology Twitter has, they can immediately stop these accounts, but they have done nothing to stop the dissemination and recruitment of lone wolf terrorists.”

Neither expert explains how speech can so easily be determined to be terrorism or how blanket filtering/account blocking wouldn't result in a sizable amount of collateral damage to innocent users. Mr. Farid, in particular, seems to believe sussing out terrorist-supporting speech should be as easy as flagging known child porn with distinct hashes. A tweet isn't a JPEG and speech can't be as easily determined to be harmful. It's easier said than done, but the argument here is the same as the FBI's argument in respect to "solving" the encryption "problem:" the smart people could figure this out. They're just not trying.

Altman's suggestion is even worse: just prevent Twitter accounts from being created that use any part of the handle of a previously-blocked account.

When an account is taken down by a Defendant, assuredly all such names are tracked by Defendants. It would be trivial to detect names that appear to have the same name root with a numerical suffix which is incremented. By limiting the ability to simply create a new account by incrementing a numerical suffix to one which has been deleted, this will disrupt the ability of individuals and organizations from using Defendants networks as an instrument for conducting terrorist operations.

It's all so easy when you're in the business of holding US-based tech companies responsible for acts of worldwide terrorism. First, this solves nothing. If the incremental option goes away, new accounts will be created with other names. Pretty soon, a great deal of innocuous handles will be auto-flagged by the system, preventing users from creating accounts with the handle they'd prefer -- including users who've never had anything to do with terrorism. Seriously a stupid idea, especially since the Twitter handle used in the example is "DriftOne" -- a completely innocuous handle the plaintiffs would like to see treated as inherently suspicious.

And thank your various gods this attorney isn't an elected official, law enforcement officer, or holding a supervisory role at an intelligence agency. Because this assertion would be less ridiculous and more frightening if delivered by any of the above:

Sending out large numbers of requests to connect with friends/followers from a newly created account is also suspicious activity. As shown in the “DriftOne” example above, it is clear that this individual must be keeping track of those previously connected. When an account is taken down and then re-established, the individual then uses an automated method to send out requests to all those members previously connected. Thus, accounts for ISIS and others can quickly reconstitute after being deleted. Such activity is suspicious on its face.

We've seen a lot of ridiculous lawsuits fired off in the wake of tragedies, but this one appears to be the worst one yet.

The lawsuit asks the court to sidestep Section 230 and order private companies to start restricting speech on their platforms. That's censorship and that's basically what the plaintiffs want -- along with fees, damages, etc. The lawsuit asks for an order finding that the named companies are violating the Anti-Terrorism Act and "grant other and further relief as justice requires."

The lawsuit's allegations are no more sound than the assertion of the Congressman quoted in support of the plaintiffs' extremely novel legal theories:

“Terrorists are using Twitter,” Rep. Poe added, and “[i]t seems like it’s a violation of the law.”

This basically sums up the lawsuit's allegations: this all "seems" wrong and the court needs to fix it. The shooting in Orlando was horrific and tragic. But this effort doesn't fix anything and asks for the government to step in and hold companies accountable for third-party postings under terrorism laws. Not only that, but it encourages the government to pressure these companies into proactive censorship based on little more than some half-baked assumptions about how the platforms work and what tech fixes they could conceivably apply with minimal collateral damage.

Filed Under: cda 230, isis, keith altman, material support for terrorism, omar mateen, orlando, pulse nightclub, section 230, terrorism, victims
Companies: facebook, google, twitter

FBI Vacuums Up Local Law Enforcement Documents To Block Open Records Requests About Orlando Shooting

from the PROPERTY-OF-THE-FBI-stamp-ink-running-dangerously-low dept

The FBI has decided to insert itself into another public records battle. The agency has long been known to cc: itself to public records requests for Stingray documents, but this time it's claiming any requests for local law enforcement documents related to the Orlando nightclub shooting need to be routed through it.

A June 20 letter from the FBI, attached to the City or Orlando's lawsuit over withholding 911 calls and other records from 25 media outlets including the Orlando Sentinel, was also sent to the Seminole County Sheriff's Office with instructions pertaining to how they should respond to records requests.

The letter requests that agencies deny inquiries and directs departments to "immediately notify the FBI of any requests your agency received" so "the FBI can seek to prevent disclosure through appropriate channels, as necessary."

Notice the FBI says "prevent disclosure," rather than, say, "assist in determining which documents can be released." The letter [PDF] claims that all records generated by local law enforcement agencies are now the FBI's by proxy and that the "investigative documents" exemption [Exemption 7(A)] prevents the release of all documents related to the shooting.

The FBI flat-out states all documents are belong to it.

The FBI considers information obtained from state and local law enforcement agencies in furtherance of its investigation to be evidence, or potential evidence.

Presto! Instant blanket exemption from disclosure at both federal and state level. The FBI takes care to point out which Florida Sunshine Law exemption local agencies can use to withhold documents from requesters.

There's significant public interest in these documents, especially those related to EMS/police response to emergency calls. This obviously conflicts with the FBI's determination that its ongoing investigation -- which now apparently contains every document created by every responding law enforcement agency in Florida -- should preempt any and all requests for documents via Florida open records laws.

Not for nothing have there been several efforts mounted to alter blanket exemptions like the one the FBI is using to insert itself into local level records requests. Unfortunately, it's very likely the FBI's wielding of this "open investigation" exemption will be granted deference by the federal court currently presiding over an open records lawsuit between the Orlando Sentinel and the City of Orlando, even though this fight never should have included a federal agency conducting its own concurrent investigation of the mass shooting.

Filed Under: fbi, foia, imsi catcher, law enforcement, orlando, public records, stingray

Facebook Still Deleting Non-Offensive Posts For Being Offensive

from the the-best-defense dept

Another day, another example of Facebook's attempt at applying automated morality going poorly. For a site designed for little else beyond expressive speech, I suppose some erroneous applications of any kind of puritanism would go awry. Perhaps then you might have forgiven Facebook's mistaking a children's illustration for man-horse-fucking, or the algorithm's inability to recognize satire.

But you would think that, in the wake of the tragic shooting that occurred at a nightclub in Orlando, one member of the LGBT community's perfectly cogent and innocuous rant wouldn't be gobbled up the by censor algorithm as being offensive. Here is the author's tweet complaining about its removal (twice), including a screenshot of the text, so that you can get an idea of what was taken down.

After removing the post a second time, Facebook apparently banned the author from doing anything else on the site for three days. That notice pretty clearly implies that the author had shared some kind of child pornography. Which wasn't true. The entirety of the post was text, imploring society to start to treat the LGBT community with a greater deal of respect. And, whatever your thoughts on the text of that post might be, certainly it is not something that should have been taken down as offensive, nevermind as any kind of pornography.

Look, I poke fun at Facebook over its morality attempts because it's funny, but I don't expect the site to be perfect. Still, when the censors are gobbling up legitimate viewpoints basically devoid of anything offensive as offensive, then perhaps we're better off with no filters at all.

Filed Under: censorship, morality police, orlando
Companies: facebook

Both Trump And Clinton Suggest Expanding Mass Surveillance, Bogus Watch Lists After Attack In Orlando

from the because-of-course dept

The script for what to do following a tragedy like the one in Orlando over the weekend is now quite clear: politicians want to appear "serious" about the issue, and thus they say stuff to appease people, even if what they say makes no sense. There was a lot of senseless rhetoric going around, of course, and we'll leave the usual debates about issues we don't cover on Techdirt to lots of other sites. But an issue we do cover is surveillance and bogus ideas like "watch lists" where a mere accusation leads to basic rights being taken away. And, unfortunately, it appears that both major Presidential candidates are advocating for greater surveillance and denial of civil liberties as a response to someone shooting up a nightclub and killing dozens of people.

Clinton's plan? Expand the "terrorist watch lists" despite the fact that there are hundreds of thousands of people who appear to be on the list for no reason at all, and whose lives are basically a living hell because of it. No matter, Clinton says let's expand it:

"We need to look carefully at this," she said. "Should we have a broader database? If someone comes to the attention of the FBI not once, but three times, that suggests that law enforcement needs to know, that people need to be more aware."

Meanwhile, Trump, beyond the much publicized and repeated plan to stop anyone who is a Muslim from immigrating to the country (even though the shooter was born here), also encouraged a much broader version of the already idiotic "see something, say something" campaign:

He also said Americans need to be willing to call the authorities when they see friends, family and neighbors performing suspicious activities.

Chelsea Manning, who is serving an insanely long sentence for leaking documents, wrote a thoughtful and insightful piece arguing for why we should not toss away our civil liberties in response to these attacks.

We must grieve and mourn and support each other, but in our grief and outrage we must resist any temptations to let this attack – or any attack – trigger anti-Muslim foreign policy, attacks on our civil liberties or as an excuse to descend into xenophobia and Islamophobia.

However, an attack like this is carefully planned and executed to maximize attention by inflaming the passions of a helpless public. Because of this, the response can be more dangerous than the attack. The refrains of “safety and security” have, for many years, been used as a tool by the powerful to justify curtailing civil liberties and emboldening backlash against immigrants, Muslim people and others.

Not for the first time, someone locked up on a questionable basis is making a lot more sense, and sounding a lot closer to the ideals of America, than either of people running to lead the country.

Filed Under: attacks, donald trump, hillary clinton, orlando, politics, see something say something, surveillance, terrorist watch list

Orlando Police Chief Threatens Critical Blogger, Saying Truth Isn't A Defense

from the oh-really? dept

We've seen plenty of government officials get upset about various things critics have said about them, and Tim writes in to let us know that down in Orlando, Florida, the local police chief is threatening to sue a web critic who put up a site highlighting how the chief had her gun stolen from her car, and then that news was kept secret for a while. While there are some complications here (the site the blogger is using is the chief's name, ValDemings.com, for example), it's hard to see how there's any defamation here at all, despite the Chief's claims. She does claim that he portrayed the situation in "false light," but as the article notes, the Florida Supreme Court recently ruled that "false light isn't a legitimate cause of action and has the potential to chill free speech."

But what may be most scary is the following quote from Demings' attorney:

"Truth is not always a defense. I hope he [Harris] gets himself a really good lawyer."

While it's true that some have been trying to push the boundaries of libel law to get rid of "truth" as an absolute defense, that troubles most people, and it's hardly common. Of course, in the meantime, in trying to shut up this blogger, Demings seems to be doing a great job kicking up a lot of attention about the fact she lost her gun...

Filed Under: bloggers, defamation, orlando, truth, val demings